Sandbox Wrapped","à":"to","À DÉCOUVRIR":"DISCOVER","à l'année":"to year","À propos du projet":"About the project","AA-Robotics teste des modèles de langage aux commandes d'un bras robotique":"AA-Robotics tests language models controlling a robotic arm","Ablehnen":"Reject","Accept":"Accept","Accepter":"Accept","Aceptar":"Accept","Action absente.":"Missing action.","Administrer un serveur Rust depuis le chat.":"Manage a Rust server from chat.","Afficher":"Show","Affiner la chronologie":"Refine the timeline","Agents API ajoute l'utilisation de l'ordinateur":"Agents API adds computer use","Agrandir":"Enlarge","ai":"ai","AI co-scientist utilise plusieurs agents basés sur Gemini 2.0 pour proposer, discuter et améliorer des hypothèses scientifiques.":"AI co-scientist uses several Gemini 2.0-based agents to propose, discuss and improve scientific hypotheses.","Ai2 présente une nouvelle version de son émulateur climatique qui produit des champs de pluie, de vent et de pression à haute résolution.":"Ai2 introduces a new version of its climate emulator producing high-resolution rainfall, wind and pressure fields.","Aigora compare les réponses des modèles de langage au questionnaire politique Politiscales. Pour chaque modèle, le projet publie les réponses, les scores par axe et le prompt employé.":"Aigora compares language models' answers to the Politiscales political questionnaire. For each model, the project publishes the answers, scores by axis and prompt used.","Aka":"Aka","Akasha est un assistant IA qui fonctionne autour d'un service restant actif sur la machine. Il dispose d'une interface desktop et terminal, d'une mémoire et de tâches planifiées.":"Akasha is an AI assistant built around a service that keeps running on the machine. It provides desktop and terminal interfaces, memory and scheduled tasks.","Akasha Model évalue une liste d'options et renvoie des distributions de probabilités en un passage. Les questions peuvent prendre la forme d'un choix, d'un score ou d'une condition.":"Akasha Model evaluates a list of options and returns probability distributions in a single pass. Questions can take the form of a choice, score or condition.","Akasha OS réunit conversations, modèles, mémoire et outils dans un environnement local. Les actions des agents passent par des capacités explicites. Les sessions persistantes, notes et tâches restent réunies dans la même application.":"Akasha OS brings conversations, models, memory and tools together in a local environment. Agent actions use explicit capabilities. Persistent sessions, notes and tasks stay in the same application.","Akzeptieren":"Accept","Aletheia explore des problèmes de recherche mathématique":"Aletheia explores mathematical research problems","AlexNet classe les images avec un réseau entraîné sur GPU":"AlexNet classifies images with a GPU-trained network","AlexNet montre les performances d'un grand réseau convolutif entraîné sur GPU pour classifier les images d'ImageNet.":"AlexNet demonstrates the performance of a large GPU-trained convolutional network for classifying ImageNet images.","Alibaba lance Qwen3.8-Max sur QwenCloud pour le code, la recherche et les tâches longues.":"Alibaba launches Qwen3.8-Max on QwenCloud for coding, research and long tasks.","alibaba qwen":"alibaba qwen","AlphaDev découvre de nouveaux algorithmes de tri":"AlphaDev discovers new sorting algorithms","AlphaDev utilise l'apprentissage par renforcement pour chercher des séquences d'instructions qui accélèrent le tri de données.":"AlphaDev uses reinforcement learning to search for instruction sequences that speed up data sorting.","AlphaEvolve explore la découverte d'algorithmes":"AlphaEvolve explores algorithm discovery","AlphaFold 2 prédit la structure des protéines à CASP14":"AlphaFold 2 predicts protein structures at CASP14","AlphaFold 3 étend la prédiction de structure aux interactions entre protéines, ADN, ARN et petites molécules.":"AlphaFold 3 extends structure prediction to interactions between proteins, DNA, RNA and small molecules.","AlphaFold 3 prédit les interactions moléculaires":"AlphaFold 3 predicts molecular interactions","AlphaFold ouvre plus de 200 millions de structures prédites":"AlphaFold opens over 200 million predicted structures","AlphaGenome prédit les effets de variations de l'ADN":"AlphaGenome predicts the effects of DNA variations","AlphaGeometry associe réseaux neuronaux et déduction":"AlphaGeometry combines neural networks and deduction","AlphaGo conclut sa victoire face à Lee Sedol":"AlphaGo completes its victory against Lee Sedol","AlphaGo termine la rencontre de Séoul avec quatre victoires contre une face au grand maître de go.":"AlphaGo finishes the Seoul match with four wins to one against the Go grandmaster.","AlphaMissense classe les effets probables de mutations qui changent un acide aminé dans une protéine humaine.":"AlphaMissense classifies the likely effects of mutations that change an amino acid in a human protein.","AlphaMissense prédit les effets de mutations génétiques":"AlphaMissense predicts the effects of genetic mutations","AlphaProof et AlphaGeometry 2 atteignent le niveau argent à l'IMO":"AlphaProof and AlphaGeometry 2 reach silver-medal level at the IMO","AlphaProteo conçoit des protéines destinées à se lier à des cibles moléculaires pour la recherche en biologie et en santé.":"AlphaProteo designs proteins that bind to molecular targets for biology and health research.","AlphaProteo génère des protéines de liaison":"AlphaProteo generates binding proteins","AlphaTensor découvre des algorithmes de multiplication de matrices":"AlphaTensor discovers matrix multiplication algorithms","AlphaTensor trouve des algorithmes qui réduisent le nombre de multiplications pour certaines tailles de matrices.":"AlphaTensor finds algorithms that reduce the number of multiplications for certain matrix sizes.","AlphaZero apprend les échecs, le shogi et le go":"AlphaZero learns chess, shogi and Go","Animations activées":"Animations enabled","Animations réduites":"Reduced motion","Année":"Year","Année de début":"Start year","Année de fin":"End year","ANNÉE EN COURS":"CURRENT YEAR","Année en cours, recherche vérifiée le":"Current year, research checked on","Annonce":"Announcement","Annonce d'AlphaFold 3":"AlphaFold 3 announcement","Annonce d'InstructGPT, OpenAI":"InstructGPT announcement, OpenAI","Annonce d'Operator":"Operator announcement","Annonce de BLOOM par BigScience":"BLOOM announcement by BigScience","Annonce de Claude 3":"Claude 3 announcement","Annonce de Claude Fable 5 et Mythos 5":"Claude Fable 5 and Mythos 5 announcement","Annonce de CLIP, OpenAI":"CLIP announcement, OpenAI","Annonce de CUDA 1.0, NVIDIA":"CUDA 1.0 announcement, NVIDIA","Annonce de DALL-E, OpenAI":"DALL-E announcement, OpenAI","Annonce de Gemini 1.5":"Gemini 1.5 announcement","Annonce de GPT-4":"GPT-4 announcement","Annonce de GPT-4o":"GPT-4o announcement","Annonce de GPT-6 Sol et Luna":"GPT-6 Sol and Luna announcement","Annonce de Grok Voice 3.0":"Grok Voice 3.0 announcement","Annonce de l'abonnement X":"X subscription announcement","Annonce de l'API AlphaGenome":"AlphaGenome API announcement","Annonce de l'API, OpenAI":"API announcement, OpenAI","Annonce de l'étude sur l'halicine, MIT":"Halicin study announcement, MIT","Annonce de l'iPhone 4S, Apple":"iPhone 4S announcement, Apple","Annonce de la Cyber Mission":"Cyber Mission announcement","Annonce de la préversion, GitHub":"Preview announcement, GitHub","Annonce de la sortie publique, Stability AI":"Public release announcement, Stability AI","Annonce de LaMDA, Google":"LaMDA announcement, Google","Annonce de lancement de Grok Bot":"Grok Bot launch announcement","Annonce de LLaMA":"LLaMA announcement","Annonce de Mistral Large 4":"Mistral Large 4 announcement","Annonce de PaLM, Google Research":"PaLM announcement, Google Research","Annonce de Team Bots":"Team Bots announcement","Annonce de Veo 3 et Imagen 4":"Veo 3 and Imagen 4 announcement","Annonce de Whisper, OpenAI":"Whisper announcement, OpenAI","Annonce des missions et du mode PvP":"Missions and PvP mode announcement","Annonce des résultats CASP14, DeepMind":"CASP14 results announcement, DeepMind","Annonce développeurs de Gemini 4 Argon":"Gemini 4 Argon developer announcement","Annonce du corpus mathématique":"Mathematical corpus announcement","Annonce du premier GPT, OpenAI":"First GPT announcement, OpenAI","Annonce et accès à Llama 2":"Llama 2 announcement and access","Annonce initiale de ChatGPT, OpenAI":"Initial ChatGPT announcement, OpenAI","Annonce initiale de GPT-2":"Initial GPT-2 announcement","Annonce initiale de Muse":"Initial Muse announcement","Annonce officielle de Dots":"Official Dots announcement","Annonce officielle de Grok 4.7":"Official Grok 4.7 announcement","Annonce officielle de Muse Wonder":"Official Muse Wonder announcement","Annonce officielle Qwen3.8-Max":"Official Qwen3.8-Max announcement","Annonce originale de WaveNet, DeepMind":"Original WaveNet announcement, DeepMind","annonces de modèles":"model announcements","annonces de modèles.

1:"$Sreact.fragment" 2:I[20631,["/_next/static/immutable/chunks/357cxq-i-o0b8.js"],"default"] 3:I[12001,["/_next/static/immutable/chunks/357cxq-i-o0b8.js"],"default"] 6:I[20227,["/_next/static/immutable/chunks/357cxq-i-o0b8.js"],"OutletBoundary"] 7:"$Sreact.suspense" a:I[20227,["/_next/static/immutable/chunks/357cxq-i-o0b8.js"],"ViewportBoundary"] c:I[20227,["/_next/static/immutable/chunks/357cxq-i-o0b8.js"],"MetadataBoundary"] e:I[38906,[],"default",1] :HL["/_next/static/immutable/chunks/2qqr5qbjg156-.css","style"] :HL["/_next/static/immutable/chunks/3yue6s5rbzt-m.css","style"] :HL["/_next/static/immutable/chunks/0uj461e_64ipx.css","style"] :HL["/_next/static/immutable/media/archivo_latin_wdth_normal-s.p.3b1_7etm2gpf5.woff2","font",{"crossOrigin":"","type":"font/woff2"}] :HL["/_next/static/immutable/media/newsreader_latin_wght_normal-s.p.1tkuwlnjs9jrl.woff2","font",{"crossOrigin":"","type":"font/woff2"}] :HL["/_next/static/immutable/chunks/3ddqhgx52-4i5.css","style"] :HL["/_next/static/immutable/chunks/145-l64j47n-s.css","style"] :HL["/_next/static/immutable/chunks/3fsjm8utimwop.css","style"] 9:X 0:{"P":null,"c":["","en","charts","history","2026-W41"],"q":"","i":false,"f":[[["",{"children":[["locale","en","d",["about","admin","api","badges","charts","confidentialite","episodes","feed.xml","latest","models","moves","nos-projets","podcast","projects","search","stories","time-machine"]],{"children":[["path","charts/history/2026-W41","oc",[]],{"children":["__PAGE__",{},"$undefined","$undefined",4096]},"$undefined","$undefined",4096]},"$undefined","$undefined",4112]},"$undefined","$undefined",4112],[["$","$1","c",{"children":[null,["$","$L2",null,{"parallelRouterKey":"children","error":"$undefined","errorStyles":"$undefined","errorScripts":"$undefined","template":["$","$L3",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":[[["$","title",null,{"children":"404: This page could not be found."}],["$","div",null,{"style":{"fontFamily":"system-ui,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif,\"Apple Color Emoji\",\"Segoe UI Emoji\"","height":"100vh","textAlign":"center","display":"flex","flexDirection":"column","alignItems":"center","justifyContent":"center"},"children":["$","div",null,{"children":[["$","style",null,{"dangerouslySetInnerHTML":{"__html":"body{color:#000;background:#fff;margin:0}.next-error-h1{border-right:1px solid rgba(0,0,0,.3)}@media (prefers-color-scheme:dark){body{color:#fff;background:#000}.next-error-h1{border-right:1px solid rgba(255,255,255,.3)}}"}}],["$","h1",null,{"className":"next-error-h1","style":{"display":"inline-block","margin":"0 20px 0 0","padding":"0 23px 0 0","fontSize":24,"fontWeight":500,"verticalAlign":"top","lineHeight":"49px"},"children":404}],["$","div",null,{"style":{"display":"inline-block"},"children":["$","h2",null,{"style":{"fontSize":14,"fontWeight":400,"lineHeight":"49px","margin":0},"children":"This page could not be found."}]}]]}]}]],[]],"forbidden":"$undefined","unauthorized":"$undefined"}]]}],{"children":[["$","$1","c",{"children":[[["$","link","0",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/2qqr5qbjg156-.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","link","1",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/3yue6s5rbzt-m.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","link","2",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/0uj461e_64ipx.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","script","script-0",{"src":"/_next/static/immutable/chunks/3oguwynlpfg8i.js","async":true,"nonce":"$undefined"}],["$","script","script-1",{"src":"/_next/static/immutable/chunks/0xzowx1_cz2pp.js","async":true,"nonce":"$undefined"}],["$","script","script-2",{"src":"/_next/static/immutable/chunks/06gfbzk5l_5yy.js","async":true,"nonce":"$undefined"}],["$","script","script-3",{"src":"/_next/static/immutable/chunks/24mhef7zkfs6-.js","async":true,"nonce":"$undefined"}]],"$L4"]}],{"children":[["$","$1","c",{"children":[null,["$","$L2",null,{"parallelRouterKey":"children","error":"$undefined","errorStyles":"$undefined","errorScripts":"$undefined","template":["$","$L3",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":"$undefined","forbidden":"$undefined","unauthorized":"$undefined"}]]}],{"children":[["$","$1","c",{"children":["$L5",[["$","link","0",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/3ddqhgx52-4i5.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","link","1",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/145-l64j47n-s.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","link","2",{"rel":"stylesheet","href":"/_next/static/immutable/chunks/3fsjm8utimwop.css","precedence":"next","crossOrigin":"$undefined","nonce":"$undefined"}],["$","script","script-0",{"src":"/_next/static/immutable/chunks/2ghckoypcz7ea.js","async":true,"nonce":"$undefined"}],["$","script","script-1",{"src":"/_next/static/immutable/chunks/07a5qibpo8z0n.js","async":true,"nonce":"$undefined"}],["$","script","script-2",{"src":"/_next/static/immutable/chunks/3iesc6gvkoa3o.js","async":true,"nonce":"$undefined"}],["$","script","script-3",{"src":"/_next/static/immutable/chunks/1maqrd5tp0foz.js","async":true,"nonce":"$undefined"}]],["$","$L6",null,{"children":["$","$7",null,{"name":"Next.MetadataOutlet","children":"$@8"}]}]]}],{},null,false,null]},null,false,"$9"]},null,false,null]},null,false,"$9"],["$","$1","h",{"children":[null,["$","$La",null,{"children":"$Lb"}],["$","div",null,{"hidden":true,"children":["$","$Lc",null,{"children":["$","$7",null,{"name":"Next.Metadata","children":"$Ld"}]}]}],["$","meta",null,{"name":"next-size-adjust","content":""}]]}],false]],"m":"$undefined","G":["$e",[]],"S":false,"h":null,"r":"$undefined","s":"$undefined","a":"$undefined","l":"$undefined","p":"$undefined","d":"$undefined","b":"nJp3ZpUmuxyHWIAAEuWId"} 9:C b:[["$","meta","0",{"charSet":"utf-8"}],["$","meta","1",{"name":"viewport","content":"width=device-width, initial-scale=1"}],["$","meta","2",{"name":"theme-color","content":"#05090d"}],["$","meta","3",{"name":"color-scheme","content":"dark"}]] f:I[60664,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"LocalizationProvider"] 10:T5c5,Four segments: Le Radar (the news), Le Build (a tool we built), Le Clash (our debate) and La Bêta (our current experiments). 🔗 RADAR SOURCES • Impulsion IA, 12 measures (Ministry of Labour) : https://travail-emploi.gouv.fr/impulsion-ia-12-mesures-pour-impulser-anticiper-et-dialoguer-face-la-transformation-du-monde-du-travail-par-lintelligence-artificielle • Émeutes lycées simulator, the vibe-coded game by Vincent Flibustier : https://emeutes-2eyfvtil.manus.game/ • Google Labs Playground : https://labs.google/playground • Mistral Large 4 : https://mistral.ai/news/mistral-large-4/ • Claude Haiku 5.5 (Anthropic) : https://www.anthropic.com/claude-haiku-5-5 • Claude Haiku 5.5 on Artificial Analysis : https://artificialanalysis.ai/changelog • Claude Startups : https://claude.com/fr/programs/startups – Startup addendum : https://www.anthropic.com/legal/startup-program-addendum – Commercial terms : https://www.anthropic.com/legal/commercial-terms – Service-specific terms (ZDR) : https://www.anthropic.com/legal/service-specific-terms • Gemini at Work 2026 (Google Cloud) : https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026 🛠️ LE BUILD : SILLAGE • The website and introduction video : https://marlburrow.github.io/sillage/ • The GitHub repository (open source, MIT) : https://github.com/MarlBurroW/sillage 🧪 LA BÊTA • Power Rush, the game by Nicolas : https://powerrush.marlburrow.dev4:["$","html",null,{"lang":"en","className":"archivo_d431af79-module__cXP8nq__variable newsreader_e97e074d-module__FfFTIG__variable newsreaderitalic_88221021-module__yltuDa__variable jetbrains_562e582b-module__T6b3mq__variable","data-scroll-behavior":"smooth","children":["$","body",null,{"children":[["$","$Lf",null,{"locale":"en","dictionary":{"Classements IA":"AI rankings","Des classements IA datés et sourcés, calculés selon des méthodes publiées.":"Dated, sourced AI rankings built with published methods.","Explorez des classements datés avec leurs candidats, leurs sources et leurs méthodes publiées.":"Explore dated rankings with their candidates, sources and published methods.","Ouvrir le classement →":"Open ranking →","Aucune édition localisée n’est encore prête. Les classements relus apparaîtront après publication de leurs données.":"No localized ranking edition is ready yet. Reviewed collections will appear after their data is published.","Semaine":"Week","{{count}} entrées classées":"{{count}} ranked entries","{{count}} éditions publiées":"{{count}} published editions","Mis à jour :":"Updated:","Aussi disponible en":"Also available in","Rang":"Rank","Nom":"Name","Score":"Score","Mouvement":"Movement","Source":"Source","Profil":"Profile","Limites":"Limitations","Preuves ({{count}})":"Evidence ({{count}})","Méthode et limites":"Methodology and limitations","NOUVEAU":"NEW","Dépôt":"Repository","Éditeur":"Publisher","Test indépendant":"Independent test","Benchmark":"Benchmark","Éditorial":"Editorial","Classements":"Rankings","Les classements":"The rankings","Dernière mise à jour :":"Last updated:","Méthode":"Methodology","Position":"Position","Sources":"Sources","Mouvements":"Movement","Ancienne page de démonstration":"Old demo page","Les chiffres, classements et avis simulés de cette page ont été retirés de l'affichage public.":"The simulated figures, rankings and opinions on this page have been removed from public display.","Consultez les classements publiés ou les épisodes du podcast.":"Browse the published rankings or podcast episodes.","Voir les classements":"View rankings","Voir les podcasts":"View podcasts","Voir l’épisode {number}":"Watch episode {number}","{name} — Podcasts vidéo, sources et classements tech":"{name} — Video podcasts, sources and tech rankings","{name} : classement GitHub, historique et avis":"{name}: GitHub ranking, history and our take","{name} : benchmarks, classement et avis":"{name}: benchmarks, ranking and our take","{name} : {ranks}. {tagline}. Historique, benchmarks, épisodes et avis de l’équipe.":"{name}: {ranks}. {tagline}. History, benchmarks, episodes and our team's take.","{name}. {tagline}. Historique, benchmarks, épisodes et avis de l’équipe.":"{name}. {tagline}. History, benchmarks, episodes and our team's take.","n°{rank} du {chart}":"no.{rank} in {chart}","{topic} : les épisodes du podcast":"{topic}: podcast episodes","Épisode {number} : {title}":"Episode {number}: {title}","{dek} Retrouvez la vidéo, les chapitres et les ressources citées.":"{dek} Find the video, chapters and cited resources.","Archives des classements · {week} · SANDBOX CHARTS":"Ranking archives · {week} · SANDBOX CHARTS","Les éditions publiées de la semaine {week}, leurs positions et leurs mouvements.":"Published editions for week {week}, their positions and movements.","Comment le {chart} est calculé : sources, critères, limites":"How {chart} is calculated: sources, criteria and limitations","Sources, pondérations, fréquence de mise à jour et limites du classement {chart}. Aucune donnée inventée : chaque chiffre renvoie à sa source.":"Sources, weights, update frequency and limitations of {chart}. Every figure links to its source.","L’édition {week} de {title}, avec ses positions et mouvements figés.":"The {week} edition of {title}, with frozen positions and movements.","L’édition {week} des projets IA en accélération sur GitHub.":"The {week} edition of AI projects gaining momentum on GitHub.","Semaine {number} : {highlights}":"Week {number}: {highlights}","Rangs hebdomadaires et mesures GitHub observées pour {name}.":"Weekly ranks and observed GitHub measurements for {name}.","Croissance":"Growth","n°{rank}":"no.{rank}","{week} : hors suivi":"{week}: not tracked","{first}, {rest}":"{first}, {rest}","{count} sem. à ce rang":"{count} weeks at this rank","{count} d’affilée":"{count} in a row","· {date}":"· {date}","S{number}":"W{number}","1er":"1","{number}e":"{number}","{week} : n°{rank}":"{week}: no.{rank}","{first} ; {rest}":"{first}; {rest}","Historique de {title}. {points}":"History of {title}. {points}","Évolution de {title}":"Evolution of {title}","Suivi, {rank} du pool (bubbling under)":"Tracked, #{rank} in the pool (bubbling under)","Première entrée au Top, à la {rank} place.":"First entry in the Top at #{rank}.","Première entrée au Top, à la {rank} place. Point fort : {dimension} ({poolRank} du pool).":"First entry in the Top at #{rank}. Strength: {dimension} (#{poolRank} in the pool).","Retour au Top à la {rank} place.":"Return to the Top at #{rank}.","Retour au Top à la {rank} place ({poolRank} du pool la semaine passée).":"Return to the Top at #{rank} (#{poolRank} in the pool last week).","+{count} place.":"+{count} place.","+{count} places.":"+{count} places.","−{count} place.":"−{count} place.","−{count} places.":"−{count} places.","+{count} place, porté par {dimension} : {score} ({delta} pts).":"+{count} place, driven by {dimension}: {score} ({delta} pts).","+{count} places, porté par {dimension} : {score} ({delta} pts).":"+{count} places, driven by {dimension}: {score} ({delta} pts).","−{count} place, freiné par {dimension} : {score} ({delta} pts).":"−{count} place, held back by {dimension}: {score} ({delta} pts).","−{count} places, freiné par {dimension} : {score} ({delta} pts).":"−{count} places, held back by {dimension}: {score} ({delta} pts).","Place inchangée ; {dimension} {score} ({delta} pts).":"Position unchanged; {dimension} {score} ({delta} pts).","Audience cumulée (vue complémentaire)":"Cumulative reach (additional view)","Classement hebdomadaire des skills publics observés sur Skills.sh et reliés à leur dépôt GitHub. Le score combine la progression des installations, celle des stars et la fraîcheur du dépôt.":"Weekly ranking of public skills observed on Skills.sh and linked to their GitHub repository. The score combines install growth, star growth and repository freshness.","Compteur cumulatif publié par Skills.sh; l'historique commence lors de l'activation de la collecte.":"Cumulative count published by Skills.sh; history begins when data collection is enabled.","Décroissance exponentielle à partir du dernier push, sur une constante de temps de 30 jours.":"Exponential decay from the latest push, with a time constant of 30 days.","Décroissance exponentielle de la date du dernier push, avec une constante de temps de 30 jours.":"Exponential decay from the date of the latest push, with a time constant of 30 days.","Différence entre les relevés GitHub espacés d'environ sept jours.":"Difference between GitHub readings taken approximately seven days apart.","Différence entre les relevés quotidiens Skills.sh espacés d'environ sept jours.":"Difference between daily Skills.sh readings taken approximately seven days apart.","Forks GitHub":"GitHub forks","Fraîcheur du dépôt":"Repository freshness","Installations cumulées":"Cumulative installs","Installations gagnées (7 j)":"Installs gained (7 d)","Nombre de forks du dépôt associé.":"Number of forks of the associated repository.","Nombre total d'étoiles du dépôt associé.":"Total stars of the associated repository.","Note principale : 50 % progression des installations, 30 % progression des stars GitHub et 20 % fraîcheur du dépôt.":"Main score: 50% install growth, 30% GitHub star growth and 20% repository freshness.","Percentile des stars gagnées depuis le relevé GitHub proche de J−7.":"Percentile of stars gained since the GitHub reading closest to seven days earlier.","Percentile du nombre d'installations gagnées depuis le relevé Skills.sh proche de J−7.":"Percentile of installs gained since the Skills.sh reading closest to seven days earlier.","Percentile du nombre total d'installations Skills.sh; cette vue complémentaire n'entre pas dans le score principal.":"Percentile of total Skills.sh installs; this additional view is excluded from the main score.","Stars GitHub":"GitHub stars","Podcasts vidéo, sources et classements tech":"Video podcasts, sources and tech rankings","{count} places gagnées":"{count} places gained","{count} places perdues":"{count} places lost","{title}, aperçu":"{title}, preview","Épisode {number}":"Episode {number}","Équipe Sandbox":"Sandbox team","{title} · {detail}":"{title} · {detail}","Agent de code":"Coding agent","Agent navigateur":"Browser agent","Automatisation":"Automation","Données pour agents":"Data for agents","Framework d'agents":"Agent framework","Génération média":"Media generation","Inférence locale":"Local inference","Inférence serveur":"Server inference","Interface":"Interface","Liste de ressources":"Resource list","Modèle open weights":"Open-weight model","Modèle propriétaire":"Proprietary model","Méthode et outils":"Method and tools","Outil de dev":"Developer tool","Plateforme d'agents":"Agent platform","category:Éditeur":"Editor","Numéro {rank}":"Rank {rank}","Momentum des cinq derniers relevés : {values}":"Momentum over the last five readings: {values}","Rang sur {count} semaines : {ranks}":"Rank over {count} weeks: {ranks}","HORS TOP {size}":"OUTSIDE TOP {size}","hors suivi":"no longer tracked","{title} : de {first} à {last}":"{title}: from {first} to {last}","1 places gagnées":"1 place gained","1 places perdues":"1 place lost","APERÇU LOCAL · Les chiffres de cet aperçu sont des fixtures de développement. En production, les classements sont calculés depuis les relevés PostgreSQL de leurs sources.":"LOCAL PREVIEW · Figures in this preview are development fixtures. In production, rankings are calculated from the PostgreSQL measurements of their sources.","- Lien (":"- Link (","— Podcasts vidéo et classements tech":"— Video podcasts and tech rankings",", inversée":", reversed",", inversée (plus bas = mieux)":", reversed (lower is better)",", semaine":", week",". Couverture :":". Coverage:",". Les chiffres, eux, sont ci-dessus.":". The figures are shown above.",". Les données viennent des sources citées ; les avis sont ceux de l’équipe, jamais mélangés.":". Figures come from the cited sources; opinions belong to the team and are kept separate.",". Les positions et mouvements sont recalculés sur ce critère.":". Positions and movements are recalculated for this criterion.","· Actuel":"· Current","· archive":"· archive","· les rangs officiels sont conservés.":"· official ranks are preserved.","· officiel":"· official","· semaine":"· week","» : un tri par critère, pas le classement officiel (":"\" : sorted by criterion, rather than the official ranking (","(nouvel onglet)":"(new tab)","(YouTube, nouvel onglet)":"(YouTube, new tab)","## Accroche":"## Headline","## Critères":"## Criteria","## Entrées et avis":"## Entries and opinions","## Fréquence":"## Frequency","## Limites":"## Limitations","## Méthode":"## Methodology","## Présentation":"## Presentation","## Sources de données":"## Data sources","#1 cette semaine":"#1 this week","← Épisode précédent":"← Previous episode","↑ ou ↓ compare la position à l’édition de la semaine précédente. NEW signifie que le projet n’y figurait pas, y compris en cas de retour. Une première édition n’a pas de comparaison. Les archives conservent la configuration, les composantes, les mesures brutes et la version du score.":"↑ or ↓ compares the position with the previous week's edition. NEW means the project was absent from that edition, including when it returns. A first edition has no comparison. Archives preserve the configuration, components, raw measurements and scoring version.","+3 places. Le plus rapide du Top 10 des poids ouverts, et il tient sur une seule machine bien équipée. C'est ce qui compte quand on héberge soi-même.":"+3 places. The fastest open-weight model in the Top 10, and it runs on one well-equipped machine. That matters when you host it yourself.","+4 places d'un coup. Google arrive avec un kit sérieux, mais on n'a pas encore vu de projet réel qui s'y tienne après le week-end de démo.":"+4 places at once. Google has released a serious toolkit, but we have yet to see a real project stick with it beyond a demo weekend.","1. Collez le lien YouTube":"1. Paste the YouTube link","2. Vérifiez et publiez":"2. Review and publish","4ᵉ après onze semaines en tête. La concurrence est arrivée, pas la lassitude : rien de dramatique.":"Fourth after eleven weeks at the top. Competition has arrived, rather than fatigue: nothing dramatic.","7 j":"7 d","À brancher":"Not connected","à l'instant":"just now","À la une":"Featured","À lire comme un cours : on comprend le protocole en trente minutes et on écrit le sien le soir même.":"Read it like a course: understand the protocol in thirty minutes, then write your own that evening.","A mis fin à trois outils et à quelques débats d'équipe sur le « bon » gestionnaire Python.":"Replaced three tools and ended several team debates about the \"right\" Python manager.","À propos":"About","À propos de l’épisode":"About this episode","À propos de sa place au":"About its position in","Accélération":"Acceleration","Accroche":"Headline","Accroche du mois":"Monthly headline","Accueil":"Home","Accueil Sandbox":"Sandbox home","Activité du dépôt":"Repository activity","Actualiser depuis YouTube":"Refresh from YouTube","Adresse email":"Email address","Âge du dépôt : 100 × exp(−âge en jours / 180).":"Repository age: 100 × exp(−age in days / 180).","Agent de développement piloté en ligne de commande : il lit le dépôt, édite des fichiers, lance des commandes et des tests, et s'étend par des skills, des hooks et des serveurs MCP.":"A command-line development agent that reads the repository, edits files, runs commands and tests, and extends through skills, hooks and MCP servers.","Agent de terminal adossé aux modèles Gemini, avec un très long contexte et une extension par serveurs MCP. Publié sous licence Apache-2.0.":"A terminal agent powered by Gemini models, with a very long context window and MCP server extensions. Released under the Apache-2.0 licence.","Agent en ligne de commande qui travaille sur votre dépôt local avec des niveaux d'autonomie réglables. Le code est publié sous licence Apache-2.0.":"A command-line agent that works on your local repository with adjustable autonomy levels. Its code is released under the Apache-2.0 licence.","Agent open source qui tourne sur votre machine, accepte plusieurs modèles et s'étend par des serveurs MCP. Un contrepoids ouvert aux agents des grands éditeurs.":"An open-source agent that runs on your machine, supports several models and extends through MCP servers. An open alternative to agents from major vendors.","Agent qui travaille dans un environnement isolé : il écrit du code, lance des commandes et navigue sur le web pour résoudre des tâches de développement. Fonctionne avec la plupart des modèles.":"An agent that works in an isolated environment: it writes code, runs commands and browses the web to solve development tasks. Works with most models.","AI Models : Gemini 3 Pro +2, Fable toujours devant":"AI Models: Gemini 3 Pro up two places, Fable still ahead","AI Models Top 20 : classement croisé des modèles IA":"AI Models Top 20: a multi-criterion AI model ranking","Ajouter ＋":"Add ＋","Ajouter au catalogue":"Add to catalogue","Ajoutez le lien YouTube avant de publier ce nouvel épisode.":"Add the YouTube link before publishing this new episode.","Aller au contenu":"Skip to content","Alternatives et voisins de classement":"Alternatives and ranking neighbours","Animateur Sandbox":"Sandbox host","Animateurs":"Hosts","Annonces Anthropic":"Anthropic announcements","Annonces Google DeepMind":"Google DeepMind announcements","Annonces OpenAI":"OpenAI announcements","Annuler les changements":"Discard changes","Aperçu du résultat de recherche":"Search result preview","APERÇU LOCAL · Ces archives utilisent les fixtures de développement.":"LOCAL PREVIEW · These archives use development fixtures.","APERÇU LOCAL · FIXTURES DE DÉVELOPPEMENT":"LOCAL PREVIEW · DEVELOPMENT FIXTURES","APERÇU LOCAL · Les chiffres de cet aperçu sont des fixtures de développement. Les éditions de production utilisent les collectes GitHub.":"LOCAL PREVIEW · Figures in this preview are development fixtures. Production editions use collected GitHub data.","API REST GitHub":"GitHub REST API","Archives":"Archives","Archives des classements par semaine":"Weekly ranking archives","Archives temporairement indisponibles":"Archives temporarily unavailable","Articles":"Articles","Artificial Analysis : comparaison de modèles":"Artificial Analysis: model comparison","Assistants et agents de code, éditeurs, revues, et la question de ce qu'on délègue.":"Coding assistants and agents, editors, reviews, and deciding what to delegate.","au Top":"in the Top","Aucun changement":"No changes","Aucun chapitre ajouté.":"No chapters added.","Aucun épisode dans ce thème pour le moment.":"No episodes in this topic yet.","Aucun épisode publié pour le moment.":"No episodes published yet.","Aucun épisode trouvé":"No episodes found","Aucun projet dans cette catégorie pour cette édition.":"No projects in this category for this edition.","Aucun résultat pour \"":"No results for \"","Aucune baisse":"No drops","Aucune baisse de rang cette semaine.":"No rank drops this week.","Aucune donnée inventée : une valeur manquante reste absente.":"No fabricated data: missing values remain absent.","Aucune entrée pour cette sélection.":"No entries match this selection.","Aucune hausse":"No gains","Aucune hausse de rang cette semaine.":"No rank gains this week.","Aucune mesure disponible pour cette capacité":"No measurements available for this capability","Aucune nouvelle entrée cette semaine.":"No new entries this week.","Aucune ressource ajoutée.":"No resources added.","Aucune ressource n’a encore été ajoutée.":"No resources have been added yet.","Audience installée : étoiles et forks cumulés.":"Established audience: cumulative stars and forks.","Audience installée : étoiles et forks. Un skill ou un agent se copie souvent plus qu'il ne se suit, donc les forks pèsent autant que les étoiles.":"Established audience: stars and forks. Skills and agents are often copied more than followed, so forks carry the same weight as stars.","Aussi cette semaine":"Also this week","Auteur":"Author","Automatisation de workflows, avec des nœuds IA, auto-hébergeable":"Self-hostable workflow automation with AI nodes","Avec":"With","Avis général":"Overall opinion","Base de départ":"Starting baseline","Bibliothèque des épisodes":"Episode library","Bibliothèque Python qui permet à un modèle de naviguer, cliquer et extraire des données dans un vrai navigateur. Un des moyens les plus directs de donner des mains à un agent.":"A Python library that lets a model browse, click and extract data in a real browser. One of the most direct ways to give an agent the ability to act.","Bleu nuit":"Midnight blue","Boîte à outils qui part d'une spécification écrite pour produire un plan, des tâches puis le code, avec l'agent de votre choix. Popularise l'idée que la spec devient le document principal.":"A toolkit that starts from a written specification to produce a plan, tasks and code with your chosen agent. It promotes the specification as the central document.","Branchée":"Connected","Bubbling under · juste derrière le Top":"Bubbling under · just outside the Top","Calcul du score":"Score calculation","calculé":"calculated","Calculé depuis les positions hebdomadaires du mois.":"Calculated from the month's weekly positions.","Calculé depuis les relevés":"Calculated from collected measurements","candidats suivis cette semaine. Les rangs au-delà du Top":"candidates tracked this week. Ranks beyond the Top","Capacité du modèle":"Model capability","caractères · utilisée aussi dans les partages et le flux RSS.":"characters · also used in shares and the RSS feed.","Carte":"Card","Carte de partage ·":"Share card ·","Catégorie":"Category","Catégorie de projet":"Project category","Ce classement sera accessible lorsque les données et sa traduction auront été publiées.":"This ranking will be available once its data and translation are published.","Ce projet n’a pas encore figuré dans une édition publiée.":"This project has not appeared in a published edition yet.","Ce qu'on a essayé cette semaine":"What we tried this week","Ce qu’on en tire":"What we learned","Ce qu’on ne sait pas faire":"What we cannot do yet","Ce que c'est":"What it is","Ce que les évaluations mesurent vraiment, et ce qu'elles laissent de côté.":"What evaluations actually measure, and what they leave out.","Ce que mesure ce classement":"What this ranking measures","Ce que mesure le classement":"What the ranking measures","Ce qui casse : retour d'expérience":"What breaks: lessons from practice","Ce qui entre dans la note":"What contributes to the score","Ces noms et liens apparaissent sur la page À propos. Laissez une URL vide si le profil n’existe pas encore.":"These names and links appear on the About page. Leave a URL empty if the profile does not exist yet.","Cette entité n’est dans aucun classement pour le moment.":"This entity is not in any ranking yet.","Cette page est introuvable.":"This page could not be found.","Cette page n’existe pas.":"This page does not exist.","cette semaine":"this week","Cette semaine ↗":"This week ↗","Cette vidéo existe déjà : épisode":"This video already exists: episode","Chaîne YouTube":"YouTube channel","Changer de langue":"Change language","Chapitres":"Chapters","Chaque chiffre renvoie à sa source.":"Every figure links to its source.","Chaque édition conserve ses positions, ses mesures et sa version du score.":"Each edition preserves its positions, measurements and scoring version.","chaque fiche commence par « ### Nom {identifiant} ».":"each entry starts with \"### Name {identifiant}\".","Chaque semaine, nos épisodes vidéo explorent les outils, les modèles et les idées qui changent le travail. Retrouvez aussi les sources citées et les classements commentés.":"Every week, our video episodes explore the tools, models and ideas changing how we work. Find the cited sources and our ranking commentary too.","Charger la vidéo YouTube":"Load YouTube video","chemin localisé absolu, en minuscules, sans slash final, paramètre ni fragment requis":"an absolute lowercase localized path without a trailing slash, query parameter or fragment is required","Choisir le thème":"Choose theme","Choisir un chart":"Choose a chart","Choisir un thème":"Choose a theme","Choisir une langue":"Choose a language","Choisir une semaine de ce classement":"Choose a week for this ranking","Cinquième semaine en tête. Excellent en agents, mais regardez le prix et la vitesse : le Global ne les compte pas, vous si.":"Fifth week at the top. Excellent for agents, but check the price and speed: the Global score leaves them out; your decision should include them.","Cité":"Mentioned","Classement":"Ranking","Classement à venir":"Ranking coming soon","Classement et mouvements":"Ranking and movements","Classement hebdomadaire de modèles croisés entre Arena, Hugging Face, OpenRouter et Artificial Analysis, avec critères activés selon les mesures réellement disponibles.":"A weekly model ranking combining Arena, Hugging Face, OpenRouter and Artificial Analysis, with criteria enabled according to the measurements actually available.","Classement hebdomadaire de modèles IA fondé sur Arena, Hugging Face, OpenRouter et Artificial Analysis.":"A weekly AI model ranking based on Arena, Hugging Face, OpenRouter and Artificial Analysis.","Classement hebdomadaire de skills publiés sur Skills.sh, croisés avec les évolutions de leurs dépôts GitHub.":"A weekly ranking of skills published on Skills.sh, cross-referenced with changes in their GitHub repositories.","Classement hebdomadaire des meilleurs modèles IA open source / open weights, avec benchmarks, licence, prix et historique.":"A weekly ranking of open-source / open-weight AI models, with benchmarks, licences, prices and history.","Classement hebdomadaire des meilleurs modèles open weights, sur les mêmes critères que le classement général des modèles IA.":"A weekly ranking of open-weight models, using the same criteria as the overall AI model ranking.","Classement hebdomadaire des projets GitHub à plus forte dynamique : stars, croissance, activité. Historique, méthodologie et commentaires de l'équipe.":"A weekly ranking of GitHub projects with the strongest momentum: stars, growth and activity. History, methodology and team commentary.","Classement hebdomadaire des projets GitHub à plus forte dynamique, toutes catégories confondues : croissance des stars, activité du dépôt et audience installée.":"A weekly ranking of GitHub projects with the strongest momentum across all categories: star growth, repository activity and established audience.","Classement hebdomadaire des skills de Skills.sh, croisés avec les données de leurs dépôts GitHub.":"A weekly ranking of Skills.sh skills, cross-referenced with their GitHub repository data.","Classements commentés":"Ranking commentary","Claude Code passe #1, et ce que ça dit des agents de code":"Claude Code takes #1: what it tells us about coding agents","Collecte : 02:00 UTC.":"Collection: 02:00 UTC.","Collecte en préparation.":"Data collection is being prepared.","Collection de serveurs MCP qui montrent comment exposer fichiers, bases de données ou API à un agent. Le point de départ pour comprendre et écrire le sien.":"A collection of MCP servers demonstrating how to expose files, databases and APIs to an agent. A starting point for understanding and writing your own.","Collection de skills : des dossiers d'instructions et de scripts que l'agent charge à la demande pour une tâche précise. Sert aussi d'exemple pour écrire les siens.":"A collection of skills: instruction and script folders an agent loads on demand for a specific task. It also provides examples for writing your own.","Collez une adresse de vidéo YouTube valide.":"Paste a valid YouTube video URL.","commencez par « # Titre du classement ».":"start with \"# Ranking title\".","Commencez par coller une adresse YouTube. L’épisode":"Start by pasting a YouTube URL. The episode","Comment ce classement est calculé →":"How this ranking is calculated →","Comment le":"How the","Comment se calcule son score ?":"How is its score calculated?","Comment se construit":"How we build","Commits (30 j)":"Commits (30 d)","Commits sur la branche par défaut au cours des 30 derniers jours.":"Commits on the default branch over the past 30 days.","Composante":"Component","Composantes et poids du score":"Score components and weights","Composantes figées de l’édition":"Edition's frozen components","Composants du score":"Score components","Compréhension d'images et de texte de niveau universitaire.":"University-level understanding of images and text.","Comprendre les images en plus du texte. Absente pour les modèles texte seul.":"Understanding images alongside text. Unavailable for text-only models.","Concours de mathématiques.":"Mathematics competition.","Concurrents":"Competitors","Conditions d'entrée":"Entry requirements","configuration initiale, avant la première édition":"initial configuration, before the first edition","contenus affichés":"displayed content","Contributeurs":"Contributors","Copie indisponible. Le lien reste affiché ci-dessous.":"Copy unavailable. The link remains displayed below.","Copie indisponible. Sélectionnez ce lien :":"Copy unavailable. Select this link:","Copier le lien":"Copy link","Créer un épisode":"Create episode","Critères":"Criteria","Critères et pondérations":"Criteria and weights","Croissance hebdo":"Weekly growth","Croissance relative":"Relative growth","D’où viennent les chiffres":"Where the figures come from","dans « Limites », commencez chaque ligne par « - ».":"under \"Limitations\", start each line with \"-\".","dans « Sources de données », utilisez « - `identifiant` — usage ».":"under \"Data sources\", use \"- `identifier` — usage\".","Dans cet épisode":"In this episode","Dans cette édition":"In this edition","Dans cette fiche":"On this page","Dans les épisodes":"In the episodes","Dans les podcasts":"In the podcasts","Date de publication":"Publication date","de présence":"of presence","Découverte : 01:00 UTC.":"Discovery: 01:00 UTC.","Découvrir l’épisode":"Watch the episode","DeepSeek en tête depuis douze semaines, Kimi qui s'accroche, Qwen qui se met à jour : où en sont vraiment les modèles ouverts face aux propriétaires ?":"DeepSeek has led for twelve weeks, Kimi is holding on and Qwen keeps updating: where do open models really stand against proprietary models?","Définition de l'Open Source AI (OSI)":"Open Source AI definition (OSI)","Départage":"Tie-breaking","dépôts · +":"repositories · +","Dernière place du Top 10. Une liste de liens est-elle un projet ? Débat ouvert, on a perdu une demi-heure dessus en studio.":"Last place in the Top 10. Is a list of links a project? Still up for debate; we spent half an hour on it in the studio.","Derniers épisodes":"Latest episodes","Derniers épisodes Sandbox":"Latest Sandbox episodes","derniers jours et ayant au moins":"past days and with at least","Des équipes d'agents spécialisés qui collaborent":"Teams of specialized agents working together","Des modèles qui agissent : outils, boucles, orchestration, et tout ce qui casse en chemin.":"Models that act: tools, loops, orchestration and everything that breaks along the way.","Description":"Description","Détails et sources":"Details and sources","Deux semaines de mesures sont nécessaires pour lire ces signaux.":"Two weeks of measurements are needed to interpret these signals.","Développement piloté par les spécifications":"Specification-driven development","Différence des forks entre les deux relevés.":"Difference in forks between the two measurements.","Différence entre le relevé du jour et celui de J−7, avec une tolérance bornée.":"Difference between today's measurement and the one from seven days ago, within a bounded tolerance.","Documentation":"Documentation","Dollars par million de tokens en entrée.":"Dollars per million input tokens.","Dollars par million de tokens en sortie.":"Dollars per million output tokens.","Donne à un agent la possibilité d'ouvrir des pages, de cliquer et de lire l'état d'une application web, en s'appuyant sur l'arbre d'accessibilité plutôt que sur des captures.":"Lets an agent open pages, click and read a web application's state through its accessibility tree rather than screenshots.","Données temporairement indisponibles":"Data temporarily unavailable","Données temporairement indisponibles.":"Data temporarily unavailable.","Du dépôt à l'édition":"From repository to edition","Du dépôt à l’édition":"From repository to edition","Du pair programming en terminal, intégré à Git":"Terminal pair programming integrated with Git","durée":"duration","Durée · secondes":"Duration · seconds","échelle linéaire":"linear scale","échelle logarithmique":"logarithmic scale","Éditeur natif écrit en Rust, pensé pour la vitesse et le travail à plusieurs, qui intègre des agents de code et des serveurs MCP.":"A native editor written in Rust, designed for speed and collaboration, with coding agents and MCP server support.","Éditeur visuel de workflows reliant des centaines de services, avec des nœuds dédiés aux modèles et aux agents. Le code est visible, mais la licence limite certains usages commerciaux.":"A visual workflow editor connecting hundreds of services, with dedicated model and agent nodes. Its code is visible, but the licence restricts some commercial uses.","Édition":"Edition","Édition : chaque jour à 03:00 UTC.":"Edition: every day at 03:00 UTC.","Édition en cours · jusqu’à":"Current edition · through","Édition indisponible":"Edition unavailable","Effacer les filtres":"Clear filters","En brouillon, le nouvel épisode reste invisible au public.":"A new draft episode remains hidden from the public.","en cours ·":"current ·","Enchaîner des actions avec des outils sans dérailler.":"Chaining actions with tools without going off track.","Enregistrer le brouillon":"Save draft","Ensemble de skills pour agents de code qui cadre le travail : clarifier, planifier, écrire les tests, puis implémenter. Intéressant pour ce qu'il dit de la discipline qu'on veut imposer à un agent.":"Skills for coding agents that structure the work: clarify, plan, write tests, then implement. Interesting for the discipline we want an agent to follow.","entité en double":"duplicate entity","Entrée au Top 10 à la 6ᵉ place. On reste prudents : on a déjà vu de grands noms disparaître en trois semaines.":"Enters the Top 10 at sixth place. We remain cautious: we have seen major names disappear in three weeks.","entrées":"entries","entrées affichées dans":"entries shown in","Entrées et avis":"Entries and opinions","Entrées et sorties":"Entries and exits","Envie d'explorer ?":"Ready to explore?","épisode":"episode","Épisode":"Episode","Épisode suivant →":"Next episode →","Épisodes à venir":"Upcoming episodes","Épisodes liés":"Related episodes","Épisodes voisins":"Nearby episodes","Erreur 404":"404 error","Essayez un autre mot ou un autre thème.":"Try another word or topic.","Essayez un mot plus court ou parcourez la bibliothèque.":"Try a shorter word or browse the library.","Essayez une autre catégorie ou une autre semaine.":"Try another category or week.","est calculé":"is calculated","et sa fiche seront préparés automatiquement.":"and its page will be prepared automatically.","était":"was","Étoiles gagnées sur les 7 derniers jours.":"Stars gained over the past seven days.","Évolution de la vitesse des stars sur deux périodes successives de sept jours. La période précédente doit être positive.":"Change in star velocity across two consecutive seven-day periods. The previous period must be positive.","Évolution du nombre de contributeurs si les deux relevés existent. Cette collecte facultative n’est pas encore connectée.":"Change in contributor count when both measurements exist. This optional collection is not connected yet.","Explore des sites et renvoie du Markdown ou des données structurées prêtes à être injectées dans un contexte. Brique fréquente des pipelines de recherche pour agents.":"Explores websites and returns Markdown or structured data ready to add to a context. A common building block in agent research pipelines.","Explorer la bibliothèque":"Explore the library","Explorer les autres thèmes":"Explore other topics","Explorer les podcasts par thème":"Explore podcasts by topic","Explorer les thèmes":"Explore topics","Explorer par thème":"Explore by topic","Explorer Sandbox":"Explore Sandbox","Explorez les autres sujets de la bibliothèque.":"Explore other subjects in the library.","Explorez les derniers classements IA mesurés.":"Explore the latest measured AI rankings.","Sandbox, le podcast vidéo sur l'IA et la tech. Retrouvez les épisodes, leurs sources et les classements commentés.":"Sandbox is a video podcast about AI and tech. Explore the episodes, their sources and the charts with commentary.","Export en cours":"Exporting","Export indisponible dans ce navigateur.":"Export unavailable in this browser.","Export indisponible.":"Export unavailable.","Export PNG indisponible.":"PNG export unavailable.","Extension d'éditeur qui propose des modifications et des commandes que vous approuvez une par une. Choisit le modèle que vous voulez, avec votre propre clé.":"An editor extension that suggests changes and commands for you to approve individually. Uses your chosen model with your own key.","Faire tourner des modèles : serveurs d'inférence, quantification, coûts.":"Running models: inference servers, quantization and costs.","Faire tourner des modèles en local en une commande":"Run local models with one command","Fenêtre de contexte":"Context window","Fermer":"Close","Fermer l'export":"Close export","Fiche":"Profile","Fiche de l’épisode":"Episode page","Fiche de lecture invalide dans l'édition.":"Invalid reading card in this edition.","Fiche mise à jour avec le relevé du":"Profile updated with the measurement from","Fil d'Ariane":"Breadcrumb","Fil d’Ariane":"Breadcrumb","Filtrer les épisodes par thème":"Filter episodes by topic","Filtrer les résultats de recherche":"Filter search results","Filtrer par type":"Filter by type","Flux RSS":"RSS feed","Fois #1":"Times at #1","Forks gagnés":"Forks gained","Format du visuel":"Image format","Fraîcheur":"Freshness","Fraîcheur du dépôt (20 %)":"Repository freshness (20 %)","Framework pour développeurs TypeScript : agents, workflows, mémoire, évaluations et observabilité dans le même outil. Un des rares à viser d'abord l'écosystème JavaScript.":"A TypeScript developer framework combining agents, workflows, memory, evaluations and observability. One of the few to prioritize the JavaScript ecosystem.","Framework Python de Google pour assembler des agents et des systèmes multi-agents, avec des outils d'évaluation et un déploiement prévu pour son cloud.":"Google's Python framework for assembling agents and multi-agent systems, with evaluation tools and deployment designed for its cloud.","Framework Python où l'on définit des rôles, des tâches et un processus d'équipe. Plus simple à prendre en main qu'un graphe explicite, au prix de moins de contrôle fin.":"A Python framework for defining roles, tasks and a team process. Easier to start with than an explicit graph, at the cost of less detailed control.","Français":"French","Fréquence":"Frequency","gagnées":"gained","Génération d'images et de vidéo par graphes de nœuds":"Image and video generation through node graphs","Générée à partir du code, donc toujours à jour":"Generated from the code, so always up to date","Gestionnaire de paquets et d'environnements Python, écrit en Rust":"Python package and environment manager written in Rust","Gestionnaires de paquets, runtimes, éditeurs, automatisation : le quotidien du développeur.":"Package managers, runtimes, editors and automation: a developer's everyday tools.","GitHub Top 10 : Claude Code passe #1":"GitHub Top 10: Claude Code takes #1","GitHub Top 10 : les projets qui montent cette semaine":"GitHub Top 10: this week's rising projects","GitHub Top 10 : vLLM, troisième semaine en tête":"GitHub Top 10: vLLM's third week at the top","GitHub Top 20, Skills, Models et Rising. Les positions, les mouvements et les projets IA à surveiller, chaque jour.":"GitHub Top 20, Skills, Models and Rising. Positions, movements and AI projects to watch, every day.","Hall of #1 · qui a tenu la tête":"Hall of #1 · who led the ranking","Héberger soi-même : ce que ça coûte vraiment":"Self-hosting: what it really costs","Historique de la méthode":"Methodology history","Historique des classements":"Ranking history","Historique en cours":"History is being collected","HISTORIQUE EN COURS DE COLLECTE":"HISTORY IS BEING COLLECTED","Historique hebdomadaire":"Weekly history","Hors du top":"Outside the top","Hors Top":"Outside the Top","HORS TOP":"OUTSIDE THE TOP","Hors Top 10 cette semaine":"Outside the Top 10 this week","Hugging Face : modèles":"Hugging Face: models","Identité":"Identity","Il sort du Top 10. Le multimodal et le très long contexte sont toujours là, le reste a vieilli.":"Drops out of the Top 10. Multimodality and the very long context are still there; the rest has aged.","Import en cours…":"Importing…","Import YouTube impossible.":"YouTube import failed.","importé(s)":"imported","Importer":"Import","Indice de présence aux meilleures places sur les semaines disponibles de la période. Les valeurs brutes affichées sont les derniers relevés disponibles.":"An index of presence near the top over the available weeks in the period. Displayed raw values are the latest available measurements.","Indisponible":"Unavailable","Inférence de LLM en C/C++, sur à peu près tout le matériel":"LLM inference in C/C++, on almost any hardware","Infra et inférence":"Infrastructure and inference","Installations Skills.sh (50 %)":"Skills.sh installs (50 %)","Interface de conversation complète (historique, documents, outils) qui se branche sur Ollama ou sur n'importe quelle API compatible. Utilisée par beaucoup d'équipes comme chat interne.":"A complete chat interface with history, documents and tools, connecting to Ollama or any compatible API. Many teams use it for internal chat.","Interface en graphe pour composer des pipelines de génération d'images, de vidéo et d'audio. Très appréciée pour la précision de contrôle et l'écosystème de nœuds communautaires.":"A graph interface for composing image, video and audio generation pipelines. Valued for precise control and its ecosystem of community nodes.","Interface visuelle pour assembler des applications à base de modèles : RAG, workflows, agents et déploiement. Une alternative low-code aux frameworks.":"A visual interface for building model-based applications: RAG, workflows, agents and deployment. A low-code alternative to frameworks.","Intro et météo de la semaine":"Introduction and this week's overview","JOUR":"DAY","Jamais classé":"Never ranked","jour). Un dépôt bloqué, exclu, archivé ou devenu un fork reste hors du classement.":"day). Blocked, excluded or archived repositories and forks remain outside the ranking.","Kit de base pour assembler des agents : instructions, outils, transferts entre agents et garde-fous. Volontairement petit, avec peu d'abstractions à apprendre.":"A basic kit for building agents: instructions, tools, handoffs between agents and guardrails. Deliberately small, with few abstractions to learn.","L'agent de code d'OpenAI, en local dans le terminal":"OpenAI's coding agent, running locally in the terminal","L'agent en ligne de commande de Google":"Google's command-line agent","L’avis de l’équipe, signé":"The team's signed opinion","L’épisode à la une utilise son propre titre et son résumé sur l’accueil.":"The featured episode uses its own title and summary on the home page.","L’équipe":"The team","L'équipe n'a pas encore ajouté sa sélection pour cette édition.":"The team has not added its selection for this edition yet.","L’intention de recherche et la méthode de classement doivent correspondre.":"Search intent must match the ranking methodology.","La bibliothèque":"The library","La bibliothèque Sandbox":"The Sandbox library","La chaîne YouTube":"The YouTube channel","La collecte de ce chart se prépare.":"Data collection for this chart is being prepared.","La collecte enregistre chaque jour les compteurs GitHub, leur date et leur source. Le lundi, après la collecte du jour, les scores produisent une édition figée. Les textes de l’équipe se publient séparément.":"Collection records GitHub counters, dates and sources every day. On Monday, after that day's collection, scores produce a frozen edition. Team commentary is published separately.","La courbe apparaîtra après plusieurs collectes. Aucun point ne sera extrapolé.":"The curve will appear after several collections. No points will be extrapolated.","La couverture mensuelle sera disponible après la première édition.":"Monthly coverage will be available after the first edition.","La découverte recherche des dépôts publics autour des sujets IA. La qualification automatique retient les dépôts non archivés, non forks, actifs dans les":"Discovery searches for public repositories around AI topics. Automatic qualification keeps non-archived repositories that are not forks and have been active in the","La dernière génération de la famille Claude 5":"The latest generation of the Claude 5 family","La documentation à jour des bibliothèques, dans le contexte de l'agent":"Up-to-date library documentation in the agent's context","La fiche sera accessible lorsque la connexion à la base sera rétablie.":"The profile will be available when the database connection is restored.","La formule, ligne par ligne":"The formula, line by line","La méthode pour repérer les projets IA en accélération sur GitHub.":"The method for identifying AI projects gaining momentum on GitHub.","La mixture d'experts multimodale de Meta":"Meta's multimodal mixture of experts","La première édition sera publiée lorsque les dépôts suivis auront assez de relevés. Chaque position reposera sur des mesures enregistrées.":"The first edition will be published when tracked repositories have enough measurements. Every position will be based on recorded data.","La première place de chaque classement, par édition.":"First place in each ranking, by edition.","La section Entrées et avis manque dans le document.":"The Entries and opinions section is missing from the document.","La tech et l'IA, épisode par épisode. La vidéo, les chapitres et tous les liens cités au même endroit.":"Tech and AI, episode by episode. Videos, chapters and every cited link in one place.","La tech mérite mieux que du bruit.":"Tech deserves more than noise.","La variante de Qwen3 spécialisée dans le code et les agents":"The Qwen3 variant specialized in coding and agents","La version rapide et économique de Gemini 3":"The fast, affordable version of Gemini 3","La vidéo de cet épisode n'est pas encore disponible.":"The video for this episode is not available yet.","Laisser un agent piloter un navigateur web":"Let an agent control a web browser","Lancement du classement, version 1 de la méthode.":"Ranking launch, version 1 of the methodology.","Lancer la recherche":"Run search","le":"the","Le 70 milliards de paramètres de Meta, très répandu":"Meta's widely used 70-billion-parameter model","Le brouillon ne modifie pas le site public avant publication.":"A draft does not change the public site until it is published.","Le chart, prêt à partager.":"The chart, ready to share.","Le chemin doit utiliser le préfixe correspondant à la langue.":"The path must use the prefix matching its language.","Le classement de la semaine":"This week's ranking","Le collage charge automatiquement les informations, chapitres et liens disponibles.":"Pasting automatically loads available information, chapters and links.","Le compromis qualité, vitesse et prix de la famille Claude":"Claude's balance of quality, speed and price","Le dépôt de référence des Agent Skills":"The reference Agent Skills repository","Le développement piloté par les spécifications, avec des agents":"Specification-driven development with agents","Le grand modèle ouvert de Mistral AI":"Mistral AI's large open model","Le kit de Google pour construire et évaluer des agents":"Google's toolkit for building and evaluating agents","Le lien a peut-être changé. Retrouvez un épisode, explorez les classements ou lancez une recherche.":"The link may have changed. Find an episode, explore the rankings or run a search.","Le meilleur rapport vitesse/prix du Top 10. Si vous n'avez pas besoin de la dernière marche de qualité, c'est ici que ça se passe.":"The Top 10's best speed-to-price ratio. If you do not need the very highest quality, start here.","Le minimum ne peut pas dépasser la taille du classement.":"The minimum cannot exceed the ranking size.","Le modèle de raisonnement qui a popularisé les poids ouverts performants":"The reasoning model that popularized high-performing open weights","Le modèle de xAI":"xAI's model","Le modèle généraliste d'OpenAI":"OpenAI's general-purpose model","Le modèle généraliste ouvert qui a précédé la série V3.2":"The open general-purpose model that preceded the V3.2 series","Le modèle haut de gamme d'Anthropic pour le code et le raisonnement":"Anthropic's high-end coding and reasoning model","Le modèle multimodal à très long contexte de Google":"Google's multimodal model with a very long context window","Le motif réflexion-action qui structure la plupart des agents.":"The reasoning-and-action pattern behind most agents.","Le mouvement en chiffres":"The movement in figures","Le moyen le plus simple d'essayer un modèle ouvert. À vingt secondes de la première réponse, c'est difficile de faire mieux.":"The simplest way to try an open model. Twenty seconds to the first answer is hard to beat.","Le pipeline actuel collecte les stars, forks, watchers, issues ouvertes et commits récents quand ils sont disponibles. Les contributeurs et releases sont conservés comme absents tant que leur collecte dédiée n’est pas branchée. Skills et Models ont leur interface, avec leurs sources réelles à connecter.":"The current pipeline collects stars, forks, watchers, open issues and recent commits when available. Contributors and releases remain absent until their dedicated collection is connected. Skills and Models have their interfaces, with real sources still to connect.","Le plus gros gain de stars de la semaine. Le vrai test : est-ce qu'on s'en sert encore dans un mois, ou est-ce qu'on parle d'un joli README ?":"The week's largest star gain. The real test: will we still use it in a month, or are we discussing an attractive README?","Le plus gros mouvement de la semaine : +2 places grâce à une révision de ses scores d'agent. À regarder de près, surtout pour le contexte long.":"The week's largest movement: two places gained after revised agent scores. Worth a closer look, especially for long contexts.","Le podcast qui teste, débat et remet les chiffres en perspective.":"The podcast that tests, debates and puts figures into perspective.","Le podium":"The podium","Le premier classement se construit.":"The first ranking is being prepared.","Le projet est-il vivant ? Commits récents et nombre de contributeurs.":"Is the project active? Recent commits and contributor count.","Le score principal croise l'Elo texte d'Arena et l'Intelligence Index d'Artificial Analysis, chacun normalisé en percentile parmi les candidats. Les autres vues ne s'activent que si une source fournit une mesure comparable.":"The main score combines Arena's text Elo and Artificial Analysis's Intelligence Index, each normalized as a percentile among candidates. Other views become available only when a source supplies a comparable measurement.","Le seul du lot publié avec ses données et son code d'entraînement.":"The only model in this group released with its training data and code.","Le sujet : le développement piloté par les spécifications":"The topic: specification-driven development","Le thème est appliqué pour cette visite, mais ne peut pas être mémorisé.":"The theme is applied for this visit but could not be saved.","Lecteur vidéo":"Video player","LECTURE SANDBOX":"SANDBOX ANALYSIS","Les archives · SANDBOX CHARTS":"Archives · SANDBOX CHARTS","Les archives apparaîtront après la publication du premier classement fondé sur les relevés collectés.":"Archives will appear after the first ranking based on collected measurements is published.","Les avis (OUR TAKE) sont séparés et signés.":"Opinions (OUR TAKE) are separate and signed.","Les champs essentiels sont prêts.":"Essential fields are ready.","Les chapitres seront ajoutés à cette fiche.":"Chapters will be added to this page.","les derniers épisodes":"the latest episodes","Les données sont momentanément indisponibles.":"Data is temporarily unavailable.","Les éditions figées apparaîtront ici après publication.":"Frozen editions will appear here after publication.","Les éditions publiées et leurs premières places, conservées avec leur méthode de calcul.":"Published editions and their first-place entries, preserved with their scoring methodology.","Les épisodes, leurs ressources et les classements IA.":"Episodes, their resources and AI rankings.","Les horodatages de la description YouTube sont importés automatiquement quand ils existent. Le temps reste modifiable en secondes.":"Timestamps from the YouTube description are imported automatically when present. Times remain editable in seconds.","Les inscriptions ne sont pas encore ouvertes.":"Sign-ups are not open yet.","Les interruptions de collecte restent visibles.":"Collection interruptions remain visible.","Les mesures de croissance deviennent des percentiles entre 0 et 100 dans le groupe éligible. Les égalités reçoivent leur rang moyen. L’activité et la fraîcheur sont déjà bornées à 100. Le score est la somme pondérée de ces composantes.":"Growth measurements become percentiles between 0 and 100 within the eligible group. Ties receive their average rank. Activity and freshness are already bounded at 100. The score is the weighted sum of these components.","Les mouvements de la semaine":"This week's movements","Les plus anciens":"Oldest first","Les plus courts":"Shortest first","Les plus récents":"Newest first","Les podcasts":"Podcasts","Les poids doivent totaliser 1.":"Weights must add up to 1.","Les poids ouverts d'OpenAI, sous Apache-2.0":"OpenAI's open weights under Apache-2.0","Les premières éditions se préparent":"The first editions are being prepared","Les prochains épisodes arrivent ici.":"Upcoming episodes will appear here.","Les profils GitHub, LinkedIn et X de Lou, Nicolas et Loïc, l’équipe Sandbox.":"The GitHub, LinkedIn and X profiles of Lou, Nicolas and Loïc, the Sandbox team.","Les projets IA dont l'accélération mesurée mérite votre attention.":"AI projects whose measured acceleration deserves your attention.","Les projets qui bougent le plus cette semaine, pas ceux qui ont le plus de stars depuis dix ans.":"The projects moving most this week, rather than those that have accumulated stars for ten years.","Les projets qui ont tenu les premières places ce mois-ci.":"The projects that held the leading places this month.","Les rangs des autres projets restent disponibles.":"Other projects' ranks remain available.","Les rangs et les scores sont calculés à partir des relevés hebdomadaires.":"Ranks and scores are calculated from weekly measurements.","Les semaines sont rattachées au mois de leur lundi.":"Weeks belong to the month of their Monday.","Les serveurs de référence du Model Context Protocol":"The reference Model Context Protocol servers","Les stars et les forks mesurent l’attention et l’adoption sur GitHub. Ils ne démontrent pas la qualité, la sécurité ni l’usage en production. Les changements de compteurs peuvent être négatifs. Les valeurs manquantes restent absentes et une panne de collecte ne détruit pas l’historique.":"Stars and forks measure attention and adoption on GitHub. They do not establish quality, security or production use. Counter changes can be negative. Missing values remain absent and a collection failure does not destroy history.","Les thèmes":"Topics","Les traitements par lots respectent les quotas GitHub et consignent leurs résultats. La première édition attend au moins":"Batch jobs respect GitHub quotas and record their results. The first edition requires at least","Les vues mensuelles et sur trois mois mesurent la présence aux meilleures places : moyenne de 100 / rang sur les semaines disponibles, avec zéro point les semaines hors classement. Elles ne moyennent pas des scores normalisés de groupes différents. Une édition datée appartient à son mois. Une semaine ISO appartient au mois de son lundi.":"Monthly and three-month views measure presence near the top: the average of 100 / rank across available weeks, with zero points for weeks outside the ranking. They do not average scores normalized across different groups. A dated edition belongs to its calendar month. An ISO week belongs to the month of its Monday.","Licence :":"Licence:","Lien · facultatif":"Link · optional","Lien à copier":"Link to copy","Lien à retrouver avant publication.":"Link to find before publication.","Lien Apple Podcasts":"Apple Podcasts link","Lien copié":"Link copied","Lien copié dans le presse-papiers":"Link copied to clipboard","Lien copié.":"Link copied.","Lien Spotify":"Spotify link","lien YouTube":"YouTube link","lien(s)":"links","lien(s) ou annexe(s)":"links or attachments","Liens cités à l’écran ou consultés pour préparer cet épisode.":"Links cited on screen or consulted to prepare this episode.","Liens cités dans la vidéo ou utilisés pour préparer l’épisode. Ils apparaissent ensemble dans la barre latérale.":"Links cited in the video or used to prepare the episode. They appear together in the sidebar.","Liens d’écoute":"Listening links","linéaire":"linear","Lire la méthode ↗":"Read the methodology ↗","Lire la méthode complète ↗":"Read the full methodology ↗","Liste de ressources communautaires pour travailler avec Claude Code. Peu de code, beaucoup d'audience : un bon thermomètre de ce que la communauté essaie.":"Community resources for working with Claude Code. Little code, a large audience: a useful indicator of what the community is trying.","Lou, Nicolas et Loïc. Les liens de l'équipe Sandbox, réunis ici.":"Lou, Nicolas and Loïc. The Sandbox team's links, together in one place.","Lou, Nicolas et Loïc. Les liens de l’équipe Sandbox, réunis ici.":"Lou, Nicolas and Loïc. The Sandbox team's links, together in one place.","Mathématiques de concours.":"Competition mathematics.","MCP : ce qu'un serveur apporte vraiment":"MCP: what a server actually adds","MCP et outils":"MCP and tools","MCP, skills, spec-kit : le nouveau kit du développeur":"MCP, skills and spec-kit: the developer's new toolkit","Même logique que le GitHub Top 10, appliquée à l'écosystème des skills, agents et outils IA. Ici, un projet se copie autant qu'il se suit : les forks pèsent donc à égalité avec les stars dans l'audience.":"The GitHub Top 10 methodology applied to skills, agents and AI tools. Projects here are copied as much as followed, so forks and stars carry equal weight in the audience component.","Mêmes données et mêmes pondérations que l'AI Models Top 10, restreintes aux modèles dont les poids sont téléchargeables. Un modèle présent dans les deux classements a le même score ; seul le rang change, parce que le pool est différent.":"The same data and weights as the AI Models Top 10, restricted to models with downloadable weights. A model in both rankings keeps the same score; only its rank changes because the candidate pool differs.","Mesure":"Measurement","Mesuré sur l'historique disponible.":"Measured over the available history.","Mesurer les projets IA qui gagnent le plus de terrain sur GitHub cette semaine.":"Measure the AI projects gaining the most ground on GitHub this week.","Mesures ·":"Measurements ·","Mesures, composants du score et sources":"Measurements, score components and sources","Métadescription de l’accueil":"Home page meta description","Méthode :":"Methodology:","Méthodologie":"Methodology","Méthodologie ·":"Methodology ·","Méthodologie →":"Methodology →","Méthodologie et sources ↗":"Methodology and sources ↗","Mettre en avant sur l’accueil":"Feature on the home page","miniature YouTube non valide":"invalid YouTube thumbnail","Mis à jour ·":"Updated ·","mis à jour chaque jour":"updated daily","Mise à jour discrète mais réelle, surtout en code. Pas de mouvement au classement, l'écart avec le #4 se réduit.":"A quiet but real update, especially for coding. No rank movement, but the gap with #4 is shrinking.","Model Context Protocol : documentation":"Model Context Protocol: documentation","Model Context Protocol, serveurs d'outils et intégrations pour agents.":"Model Context Protocol, tool servers and agent integrations.","Modèle à poids ouverts de DeepSeek, connu pour son rapport qualité/prix. Disponible chez de nombreux fournisseurs et auto-hébergeable.":"DeepSeek's open-weight model, known for its quality-to-price ratio. Available from many providers and suitable for self-hosting.","Modèle à poids ouverts de l'éditeur français, multilingue et multimodal, publié sous licence Apache-2.0.":"A multilingual and multimodal open-weight model from the French vendor, released under the Apache-2.0 licence.","Modèle à poids ouverts de MiniMax, conçu pour tourner vite et bon marché sur des boucles d'agents longues.":"MiniMax's open-weight model, designed for fast and affordable long-running agent loops.","Modèle à poids ouverts du laboratoire chinois Z.ai, régulièrement cité pour ses performances en code agentique à un prix bas.":"An open-weight model from the Chinese lab Z.ai, regularly cited for affordable performance in agentic coding.","Modèle à poids ouverts publié par OpenAI. Assez compact pour tenir sur un seul GPU haut de gamme, avec de bonnes performances en raisonnement.":"An open-weight model released by OpenAI. Compact enough for one high-end GPU, with good reasoning performance.","Modèle compact à poids ouverts, capable de tourner sur une seule carte graphique grand public. Distribué sous les conditions d'utilisation Gemma.":"A compact open-weight model that can run on one consumer graphics card. Distributed under Gemma's terms of use.","Modèle compact de Mistral AI conçu pour la latence : peu cher, rapide, suffisant pour beaucoup de tâches courantes.":"A compact Mistral AI model designed for low latency: affordable, fast and sufficient for many everyday tasks.","Modèle de 14 milliards de paramètres de Microsoft, publié sous licence MIT. Bon en raisonnement pour sa taille, limité en contexte.":"Microsoft's 14-billion-parameter model, released under the MIT licence. Strong reasoning for its size, with a limited context window.","Modèle de 253 milliards de paramètres dérivé de Llama et affiné par NVIDIA, publié avec ses poids sous licence NVIDIA Open Model.":"A 253-billion-parameter model derived from Llama and refined by NVIDIA, with weights released under the NVIDIA Open Model licence.","Modèle de l'Allen Institute for AI publié avec tout ce qu'il faut pour le reproduire. Moins performant que les grands modèles ouverts, mais le plus transparent.":"An Allen Institute for AI model released with everything needed to reproduce it. Less capable than major open models, but the most transparent.","Modèle de raisonnement publié avec ses poids, qui a montré qu'un modèle ouvert pouvait approcher les meilleurs modèles propriétaires du moment. Surpassé depuis, mais toujours une référence.":"A reasoning model released with its weights, showing that an open model could approach leading proprietary models of its time. Since surpassed, but still a reference.","Modèle dense de Meta, largement déployé et affiné par la communauté. Distribué sous une licence maison.":"Meta's dense model, widely deployed and refined by the community. Distributed under its own licence.","Modèle généraliste ouvert d'Alibaba, publié sous licence Apache-2.0, ce qui en fait un des plus simples à utiliser en entreprise.":"Alibaba's open general-purpose model under Apache-2.0, making it one of the simplest to use in business.","Modèle ouvert compact d'Alibaba, sous Apache-2.0, qui tient sur une seule carte graphique avec quantification.":"Alibaba's compact open model under Apache-2.0, fitting on one graphics card with quantization.","Modèle ouvert de Meta, multimodal et à très grand contexte. Distribué sous une licence maison qui impose des conditions d'usage.":"Meta's open multimodal model with a very large context window. Its own licence imposes conditions on use.","Modèle ouvert de Moonshot AI à très grand nombre de paramètres, entraîné pour l'usage d'outils. Un des premiers modèles ouverts à rivaliser avec les propriétaires sur les tâches d'agents.":"Moonshot AI's open model with a very large parameter count, trained for tool use. One of the first open models to rival proprietary models on agent tasks.","Modèle ouvert entraîné spécifiquement pour la programmation et l'usage d'outils, utilisable derrière la plupart des agents de code.":"An open model trained specifically for programming and tool use, compatible with most coding agents.","Modèle propriétaire d'Anthropic pensé pour un usage quotidien et les agents de code qui tournent beaucoup, avec un bon équilibre entre coût et capacités.":"Anthropic's proprietary model for everyday use and frequently running coding agents, balancing cost and capabilities.","Modèle propriétaire d'OpenAI, utilisé par ChatGPT et l'API. Souvent cité en tête sur les maths de concours et le raisonnement.":"OpenAI's proprietary model used by ChatGPT and the API. Often cited near the top for competition mathematics and reasoning.","Modèle propriétaire de Google optimisé pour la vitesse et le prix, avec un contexte très large. Le choix par défaut quand le volume compte plus que la dernière marche de qualité.":"Google's proprietary model optimized for speed and price, with a very large context window. A default choice when volume matters more than the final increment of quality.","Modèle propriétaire de Google, fort en compréhension d'images et de vidéo et sur les très longs contextes.":"Google's proprietary model, strong at understanding images and video and handling very long contexts.","Modèle propriétaire de la famille Claude 5, positionné sur les tâches longues et difficiles : développement logiciel, agents, analyse.":"A proprietary Claude 5 model for long, demanding tasks: software development, agents and analysis.","Modèle propriétaire de xAI, accessible via l'API et la plateforme X. Souvent apprécié pour sa recherche web intégrée.":"xAI's proprietary model, available through its API and X. Often valued for integrated web search.","Modèle propriétaire récent d'Anthropic. Fiche à compléter avec les informations officielles de l'éditeur.":"A recent Anthropic proprietary model. This profile needs completion using the vendor's official information.","Modèles":"Models","Modélise un agent comme un graphe de nœuds avec un état persistant, des boucles et des points de reprise. Plus bas niveau que d'autres frameworks, donc plus de contrôle.":"Models an agent as a graph of nodes with persistent state, loops and recovery points. Lower-level than other frameworks, giving more control.","Modifier l’épisode":"Edit episode","Modifier le catalogue":"Edit catalogue","Mois des archives":"Archive month","Momentum des cinq derniers relevés":"Momentum across the latest five measurements","Moteur d'inférence minimaliste et très optimisé, à la base de nombreux outils locaux. Sa quantification a rendu possible l'exécution de gros modèles sur du matériel ordinaire.":"A minimal, highly optimized inference engine powering many local tools. Its quantization enabled large models to run on ordinary hardware.","Moteur de service de modèles qui optimise le débit grâce à une gestion fine de la mémoire du cache. La référence quand on sert des modèles ouverts à plusieurs utilisateurs.":"A model-serving engine optimizing throughput through careful cache memory management. A reference for serving open models to multiple users.","Mouvements dans le classement":"Ranking movements","Mouvements et archives":"Movements and archives","n°":"no.","Navigation principale":"Main navigation","Nom de la ressource":"Resource name","Nom du site":"Site name","Nombre total d'étoiles du dépôt.":"Total repository stars.","Nombre total de forks.":"Total forks.","non évalué":"not evaluated","Normalisation":"Normalization","Normalisation et départage":"Normalization and tie-breaking","Note · facultative":"Note · optional","Note principale":"Main score","Note principale du classement : 50 % croissance, 25 % activité, 25 % audience.":"Main ranking score: 50 % growth, 25 % activity, 25 % audience.","Note principale du classement : 55 % croissance, 25 % activité, 20 % audience.":"Main ranking score: 55 % growth, 25 % activity, 20 % audience.","Notre avis":"Our take","Notre modèle par défaut pour les tâches longues, même si on a appris à lire les benchmarks en plissant les yeux.":"Our default model for long tasks, even though we have learned to read benchmarks with a critical eye.","Nouvelle entrée":"New entry","Nouvelle entrée : un dérivé de Llama affiné par NVIDIA. Intéressant, mais lourd à héberger. À essayer avant d'en tirer une conclusion.":"New entry: a Llama derivative refined by NVIDIA. Interesting, but demanding to host. Try it before drawing a conclusion.","Nouvelle entrée dans ce classement":"New entry in this ranking","Nouvelle recherche":"New search","Nouvelles stars (7 j)":"New stars (7 d)","Ollama, llama.cpp, vLLM : lequel pour quoi":"Ollama, llama.cpp and vLLM: which one for which task?","On cherche ce qui émerge, pas ce qui est installé. Un projet qui gagne 8 000 stars en une semaine passe devant un géant qui en gagne 400, sauf si ce géant est aussi très actif. Le classement mesure donc une dynamique, pas une popularité historique.":"We look for what is emerging, rather than what is established. A project gaining 8,000 stars in a week can outrank a giant gaining 400, unless that giant is also very active. The ranking measures momentum rather than historical popularity.","On passe en revue les quatre classements de la semaine 41 et on essaie de comprendre ce que chaque mouvement raconte vraiment. Puis un sujet de fond : le développement piloté par les spécifications est-il une méthode ou une mode ? Avec, comme d'habitude, la liste complète de ce qu'on a cité.":"We review the four rankings for week 41 and try to understand what each movement really tells us. Then we discuss a broader question: is specification-driven development a method or a trend? As usual, the complete list of everything we cited is included.","On s'en sert pour préparer les épisodes. Ce qui frappe, c'est moins le modèle que la boucle : il lit, il essaie, il corrige.":"We use it to prepare episodes. What stands out is less the model than the loop: it reads, tries and corrects.","Open Source AI : gpt-oss-120b +3, nouvelle entrée de NVIDIA":"Open Source AI: gpt-oss-120b up three places, a new NVIDIA entry","Open Source AI : lecture critère par critère":"Open Source AI: reading each criterion","Open Source AI Top 10 : les meilleurs modèles open weights":"Open Source AI Top 10: open-weight model rankings","Open Source Initiative : IA":"Open Source Initiative: AI","Open weights, open source, libre : de quoi parle-t-on ?":"Open weights, open source and free software: what do we mean?","Orchestrer des agents comme des graphes à état":"Orchestrating agents as stateful graphs","Organisation":"Organization","Où on en parle":"Where we discuss it","Out · sortis du Top":"Out · outside the Top","Outils":"Tools","Outils de dev":"Developer tools","Outro et questions pour la semaine prochaine":"Outro and questions for next week","Ouvrir cet épisode →":"Open this episode →","Ouvrir l'édition":"Open edition","Ouvrir la fiche ↗":"Open profile ↗","Ouvrir la fiche de":"Open the profile of","Ouvrir le chart ↗":"Open chart ↗","Page introuvable":"Page not found","Page produit":"Product page","Par thème":"By topic","Parcourez les épisodes Sandbox par sujet : IA, développement, modèles ouverts et outils.":"Browse Sandbox episodes by subject: AI, development, open models and tools.","Parcours dans le":"History in the","Part des tickets GitHub réels résolus.":"Share of real GitHub issues solved.","Partager l'épisode":"Share episode","Partager sur X ↗":"Share on X ↗","Pas de période précédente disponible":"No previous period available","Pas encore d’avis sur":"No opinion yet on","Périmètre":"Scope","Période du classement":"Ranking period","Personnes ayant au moins un commit sur la branche par défaut.":"People with at least one commit on the default branch.","Place":"Position","Place inchangée":"Position unchanged","Place inchangée.":"Position unchanged.","Plateforme pour construire des applications et agents LLM":"A platform for building LLM applications and agents","Plateformes du podcast":"Podcast platforms","Plus forte baisse":"Largest drop","Plus forte chute":"Largest drop","Plus forte hausse":"Largest gain","Plus suivi cette semaine":"No longer tracked this week","Podcast indépendant · IA · développement · open source":"Independent podcast · AI · development · open source","Podcasts : tous les épisodes":"Podcasts: all episodes","Podcasts · ressources · classements":"Podcasts · resources · rankings","Podcasts & classements IA":"Podcasts & AI rankings","Poids":"Weights","Poids fermés":"Closed weights","Poids ouverts : la remontada, graphique par graphique":"Open weights make a comeback, chart by chart","Poids ouverts uniquement. Mêmes critères que le classement général, pour comparer ce qu'on peut vraiment héberger soi-même.":"Open weights only. The same criteria as the overall ranking, to compare what you can actually host yourself.","Poids prévu":"Planned weight","Pool et identité":"Candidate pool and identity","Position inchangée":"Position unchanged","Position semaine après semaine":"Position week by week","pour \"":"for \"","Pour qui":"Who it is for","Pourquoi ça bouge":"Why it moved","Pourquoi ça compte":"Why it matters","Pourquoi ça monte":"Why it is rising","Pourquoi une révision de protocole peut déplacer un modèle de deux places.":"Why a protocol revision can move a model by two places.","Premier #1 de Claude Code ici. Pour une fois, les stars suivent l'usage : c'est un outil qu'on lance tous les jours, pas un dépôt qu'on met en favori pour « plus tard ».":"Claude Code's first #1 here. For once, stars follow usage: it is a tool we run every day, rather than a repository bookmarked for \"later\".","Premier mot du logotype":"First word of the wordmark","Première entrée":"First entry","Première génération de la série V3 de DeepSeek, à poids ouverts. Toujours très utilisée, même si les versions suivantes la dépassent.":"The first generation of DeepSeek's open-weight V3 series. Still widely used, though subsequent versions have surpassed it.","Première place pour Claude Code au GitHub Top 10, adk-python qui bondit de quatre rangs, Gemini 3 Pro qui revient sur le podium : on décortique les mouvements de la semaine 41.":"Claude Code takes first place in the GitHub Top 10, adk-python jumps four ranks and Gemini 3 Pro returns to the podium: we break down the movements of week 41.","Première semaine du classement":"First week of the ranking","Prend la tête":"Takes the lead","Prénom":"First name","Présentation":"Introduction","Présentation à compléter.":"Introduction to be completed.","Présentation courte":"Short introduction","Principes":"Principles","Prix entrée":"Input price","Prix mixte":"Blended price","Prix mixte (3 parts d'entrée pour 1 de sortie), inversé : moins cher = score plus haut.":"Blended price (three parts input to one part output), reversed: a lower price gives a higher score.","Prix sortie":"Output price","Problèmes de code récents, pour limiter la contamination des données d'entraînement.":"Recent coding problems to reduce training-data contamination.","Produits":"Products","Profils à venir":"Profiles coming soon","Progression GitHub (30 %)":"GitHub growth (30 %)","projets éligibles.":"eligible projects.","Projets et modèles":"Projects and models","Projets, licences, modèles à poids ouverts, et les gens qui maintiennent tout ça.":"Projects, licences, open-weight models and the people maintaining them.","Projets, modèles et outils":"Projects, models and tools","Publier":"Publish","Publier l’épisode":"Publish episode","Qu'est-ce qui monte dans l'IA cette semaine ?":"What is rising in AI this week?","Qualité générale (50 % Arena, 50 % Artificial Analysis)":"General quality (50 % Arena, 50 % Artificial Analysis)","Qualité globale : coding 25 %, reasoning 22 %, agents 20 %, maths 13 %, multimodal 10 %, long context 10 %. La vitesse et le prix n'y entrent pas.":"Overall quality: coding 25 %, reasoning 22 %, agents 20 %, maths 13 %, multimodal 10 %, long context 10 %. Speed and price are excluded.","Quatrième semaine en tête. Le haut du classement est stable, tout se joue derrière.":"Fourth week at the top. The leaders are stable; the movement is behind them.","Questions de sciences de niveau doctorat.":"Doctoral-level science questions.","Questions expertes : les scores restent bas, l'échelle va donc de 0 à 50 %.":"Expert questions: scores remain low, so the scale runs from 0 to 50 %.","Qui peut entrer ?":"Who can enter?","Quitter le mode 16:9":"Exit 16:9 mode","Raisonnement scientifique et questions expertes.":"Scientific reasoning and expert questions.","Rapport qualité/prix : 60 % qualité globale, 40 % score de prix.":"Value for money: 60 % overall quality, 40 % price score.","Récence du dernier push : 100 × exp(−jours depuis le push / 30).":"Latest push recency: 100 × exp(−days since push / 30).","Recherche":"Search","Recherche globale":"Global search","Rechercher":"Search","Rechercher dans les podcasts et classements":"Search podcasts and rankings","Rechercher dans les podcasts et les classements":"Search podcasts and rankings","Rechercher dans tout le site":"Search the whole site","Rechercher sur le site":"Search the site","Recherchez un épisode, une ressource, un sujet ou un classement Sandbox.":"Search for an episode, resource, topic or Sandbox ranking.","Records du classement":"Ranking records","Réessayez dans quelques instants.":"Try again in a moment.","Référencement":"SEO","Relevés hebdomadaires sur environ 30 jours":"Weekly measurements over approximately 30 days","relevés réels.":"real measurements.","Remplace pip, virtualenv et une partie de pipx et poetry avec une vitesse sans commune mesure. Devenu le réflexe par défaut pour démarrer un projet Python.":"Replaces pip, virtualenv and part of pipx and poetry with a substantial speed improvement. It has become a default choice for starting a Python project.","Repérer les projets dont la vitesse augmente avant qu’ils deviennent massifs.":"Identify projects whose velocity is increasing before they become large.","Repositories GitHub":"GitHub repositories","Résoudre de vrais tickets et écrire du code correct.":"Solving real issues and writing correct code.","Ressource":"Resource","Ressources":"Resources","résultat":"result","résumé":"summary","Résumé court":"Short summary","Résumé de la semaine":"Weekly summary","Retirer":"Remove","Retour aux charts":"Back to charts","Retour dans ce classement":"Return to this ranking","Retour dans le classement":"Return to the ranking","Retour discret. Ça ressemble à une mise à jour qui plaît aux équipes plus qu'à de la hype, ce qui est plutôt bon signe.":"A quiet return. It looks like an update valued by teams rather than hype, which is a good sign.","Retrouver plusieurs informations dans un long contexte.":"Retrieving several pieces of information from a long context.","Retrouvez bientôt les vidéos, leurs chapitres et leurs ressources.":"Videos, chapters and resources will appear here soon.","Retrouvez-nous.":"Find us online.","Retrouvez l’équipe Sandbox":"Meet the Sandbox team","Retrouvez les épisodes qui parlent des sujets qui vous intéressent.":"Find episodes about the subjects that interest you.","Retrouvez les épisodes Sandbox, leurs chapitres et les ressources mentionnées pendant les émissions.":"Find Sandbox episodes, their chapters and the resources mentioned during the shows.","Rising ajoute deux conditions : au plus":"Rising adds two requirements: at most","Rising attend une accélération mesurée sur deux périodes de sept jours. L'historique ne sera pas extrapolé.":"Rising requires measured acceleration over two seven-day periods. History will not be extrapolated.","Rouge":"Red","Saisie manuelle":"Manual entry","Saisir sans import":"Enter without importing","Saisissez au moins un mot pour rechercher dans les épisodes, les sources citées et les classements.":"Enter at least one word to search episodes, cited sources and rankings.","Saisissez une date de publication valide.":"Enter a valid publication date.","Saisissez une URL de vidéo YouTube valide.":"Enter a valid YouTube video URL.","Sandbox · le podcast":"Sandbox · the Talk Show","SANDBOX CHARTS · Méthode GitHub Top 20":"SANDBOX CHARTS · GitHub Top 20 methodology","SANDBOX CHARTS · Méthode Rising 20":"SANDBOX CHARTS · Rising 20 methodology","scores non décroissants":"scores are not decreasing","Second mot du logotype":"Second word of the wordmark","Seize semaines dans le Top, sept en tête. Il glisse à la 3ᵉ place sans avoir démérité : il se fait manger par le calendrier des sorties.":"Sixteen weeks in the Top, seven at #1. It slips to third despite performing well: new releases are overtaking it.","sélectionné(s)":"selected","sem.":"wk.","sem. au Top · pic n°":"wk. in the Top · peak #","Semaine 39":"Week 39","Semaine 39 : on démonte la pile qui s'installe autour des agents de code. Les serveurs MCP, les skills, le développement piloté par les spécifications : quelle brique vaut vraiment la peine, et laquelle est un effet de mode ?":"Week 39: we unpack the stack emerging around coding agents. MCP servers, skills and specification-driven development: which building blocks are worthwhile, and which are a passing trend?","Semaine 40":"Week 40","Semaine 40 : on regarde l'écart entre modèles ouverts et propriétaires critère par critère, on discute de ce que « open weights » veut dire et ne veut pas dire, et on teste ce que ça coûte d'héberger soi-même. Plus les mouvements des classements GitHub.":"Week 40: we compare open and proprietary models criterion by criterion, discuss what \"open weights\" means and does not mean, and test the cost of self-hosting. Plus movements in the GitHub rankings.","Semaine 41":"Week 41","Semaine du classement":"Ranking week","semaine ISO AAAA-Wnn":"ISO week YYYY-Wnn","semaines d’historique":"weeks of history","semaines dans le Top 20 · meilleur rang #":"weeks in the Top 20 · best rank #","semaines en tête":"weeks at the top","Serveur d'inférence haute performance pour LLM":"High-performance LLM inference server","Serveur MCP qui récupère la documentation et les exemples de la bonne version d'une bibliothèque pour éviter que l'agent n'invente une API périmée.":"An MCP server retrieving documentation and examples for the correct library version, to stop an agent inventing an outdated API.","Signature collective":"Collective byline","Signature des anciens contenus simulés, à remplacer lors de la publication des vrais épisodes.":"Byline for older simulated content, to replace when real episodes are published.","Signaux":"Signals","Skills : des dossiers plutôt que des prompts":"Skills: folders instead of prompts","Skills Top 10 : adk-python, superpowers et la fatigue des listes":"Skills Top 10: adk-python, superpowers and list fatigue","Skills Top 10 : la semaine d'anthropics/skills":"Skills Top 10: anthropics/skills' week","Skills Top 20 : classement des skills IA par installations et activité":"Skills Top 20: AI skills ranked by installs and activity","Skills, agents et outils IA suivis par leurs installations et l'activité de leur dépôt.":"Skills, agents and AI tools tracked through installs and repository activity.","slug en minuscules, chiffres et tirets":"a lowercase slug using letters, numbers and hyphens","Sommaire de la méthode":"Methodology contents","sont affichés pour le contexte (« bubbling under »). Relevé du":"are shown for context (\"bubbling under\"). Measurement from","Sorti du classement":"Left the ranking","Sorti du Top 10 après un pic spectaculaire. Le classique : un projet très partagé, puis la réalité des sites qui changent de structure du jour au lendemain.":"Leaves the Top 10 after a spectacular peak. A familiar pattern: a heavily shared project, followed by the reality of websites changing structure overnight.","Sorties, comparaisons, prix, vitesse : le paysage des modèles de langage.":"Releases, comparisons, prices and speed: the language-model landscape.","Source :":"Source:","SOURCE GITHUB ·":"GITHUB SOURCE ·","Sources & ressources":"Sources & resources","Sources de données":"Data sources","Sources du classement":"Ranking sources","Sources et ressources":"Sources and resources","Sources et rythme":"Sources and schedule","Sources, score, critères et limites du classement hebdomadaire GitHub.":"Sources, scores, criteria and limitations of the weekly GitHub ranking.","sous-score composite":"composite subscore","Speed et Price":"Speed and Price","Stars · relevés quotidiens":"Stars · daily measurements","stars et une accélération positive mesurée avec un deuxième relevé à J−14. Aucun historique n’est extrapolé.":"stars and positive acceleration measured with a second reading fourteen days ago. No history is extrapolated.","Stars gagnées":"Stars gained","Stars gagnées divisées par les stars du relevé précédent. Un plancher protège le score contre les très petits dénominateurs ; le pourcentage public conserve la valeur brute.":"Stars gained divided by the previous reading's stars. A floor protects the score from very small denominators; the public percentage keeps the raw value.","stars, ou en avoir gagné":"stars, or gained","stars. L’équipe peut changer leur statut de suivi et leur catégorie.":"stars. The team can change their tracking status and category.","Statut":"Status","sur la période. Il doit disposer d’un relevé à J−7 (tolérance ±":"over the period. A measurement from seven days ago is required (tolerance ±","sur une cohorte identique à la semaine précédente.":"using the same cohort as the previous week.","Sur-titre":"Eyebrow","SWE-bench : leaderboards":"SWE-bench: leaderboards","Tâches d'agent dans un terminal.":"Agent tasks in a terminal.","Taille de la fenêtre et capacité à y retrouver l'information.":"Window size and ability to retrieve information within it.","Taille maximale du contexte, en milliers de tokens.":"Maximum context size, in thousands of tokens.","Taille Models invalide.":"Invalid Models size.","Taille Skills invalide.":"Invalid Skills size.","Télécharge et exécute des modèles sur votre machine, avec une API locale compatible avec la plupart des clients. La porte d'entrée la plus simple vers les modèles ouverts.":"Downloads and runs models on your machine, with a local API compatible with most clients. The simplest entry point to open models.","Télécharger l’image":"Download image","Télécharger le PNG":"Download PNG","Temps · secondes":"Time · seconds","Textes de marque":"Brand copy","Thème":"Theme","Thèmes":"Topics","Thread sur les agents « qui font semblant de tester »":"Thread about agents that \"pretend to test\"","Titre":"Title","Titre de l’épisode":"Episode title","Titre, sujet ou ressource":"Title, topic or resource","Tokens générés par seconde (médiane des fournisseurs).":"Tokens generated per second (median across providers).","Toujours le roi des maths de concours. Mais ça fait longtemps que ce n'est plus ce qu'on lui demande au quotidien.":"Still the leader in competition mathematics. But it has been a long time since that was our main everyday requirement.","Tous les épisodes":"All episodes","Tous les liens":"All links","Tous les mois":"Every month","Tous les podcasts":"All podcasts","Tous les podcasts.":"All podcasts.","Tous les sous-scores et mesures":"All subscores and measurements","Tous les thèmes":"All topics","Tout":"All","Tout ·":"All ·","Tout voir":"View all","Toutes les catégories":"All categories","Transformer des sites web en données propres pour LLM":"Turning websites into clean data for LLMs","Transparence":"Transparency","Treize semaines en tête. Lent chez la plupart des fournisseurs, mais le prix fait oublier la latence pour les usages en batch.":"Thirteen weeks at the top. Slow with most providers, but the price makes latency easier to accept for batch use.","Trier par":"Sort by","Trier par critère":"Sort by criterion","Trois idées en même temps : donner des outils aux agents, leur apprendre des méthodes, écrire la spec avant le code. Ce qui marche, ce qui est du bruit.":"Three ideas at once: giving agents tools, teaching them methods and writing the specification before the code. What works, and what is noise.","Trois semaines en tête, puis un passage de témoin. Le projet le plus sous-estimé du classement : personne n'en parle, tout le monde en dépend.":"Three weeks at the top before handing over the lead. The ranking's most underestimated project: nobody talks about it, everybody depends on it.","Troisième semaine à la première place. Ce n'est plus une surprise, c'est une habitude.":"Third week in first place. No longer a surprise; now a habit.","Troisième semaine en tête. L'idée est simple, un dossier et un fichier d'instructions, et c'est justement pour ça qu'elle se copie partout.":"Third week at the top. The idea is simple, a folder and an instruction file, which is precisely why it is copied everywhere.","trouvé":"found","Trouver un épisode":"Find an episode","Un agent de code dans VS Code, avec validation à chaque étape":"A coding agent in VS Code, with approval at every step","Un agent de code qui vit dans votre terminal":"A coding agent that lives in your terminal","Un agent de développement logiciel open source":"An open-source software development agent","Un agent local et extensible, avec support MCP":"An extensible local agent with MCP support","Un dépôt suivi doit avoir au moins":"A tracked repository must have at least","Un dérivé de Llama optimisé par NVIDIA pour le raisonnement":"A Llama derivative optimized by NVIDIA for reasoning","Un des premiers assistants de code en terminal : il édite vos fichiers et crée un commit par changement, ce qui rend chaque modification relisable et réversible.":"One of the first terminal coding assistants: it edits your files and creates a commit for each change, making modifications reviewable and reversible.","Un des projets qui ont le plus changé ce qui est possible sur du matériel ordinaire. Peu de bruit, énormément d'effet.":"One of the projects that most changed what ordinary hardware can do. Little noise, a substantial effect.","Un éditeur de code rapide et collaboratif, avec agents intégrés":"A fast, collaborative code editor with integrated agents","Un épisode à retrouver, un outil cité, un sujet à explorer.":"An episode to find, a cited tool or a topic to explore.","Un framework TypeScript pour agents et workflows IA":"A TypeScript framework for AI agents and workflows","Un framework TypeScript qui monte vite dans un écosystème d'agents surtout écrit en Python. On attend de voir si c'est de l'élan ou un effet de mode.":"A TypeScript framework growing quickly in an agent ecosystem mostly written in Python. We are waiting to see whether this is lasting momentum or a trend.","Un modèle dense de 32 milliards de paramètres, facile à héberger":"A dense 32-billion-parameter model, easy to host","Un modèle entièrement ouvert : poids, données et code d'entraînement":"A fully open model: weights, data and training code","Un modèle généraliste ouvert, à très bas coût d'inférence":"An open general-purpose model with very low inference costs","Un modèle n'est pas « le meilleur ». Il est meilleur sur quelque chose. Changez de critère, le classement change.":"A model is not simply \"the best\". It performs better at something. Change the criterion and the ranking changes.","Un modèle ouvert généraliste et agentique de Z.ai":"Z.ai's open general-purpose and agentic model","Un modèle ouvert rapide, orienté code et agents":"A fast open model focused on coding and agents","Un petit modèle ouvert et multimodal de Google":"Google's small open multimodal model","Un petit modèle ouvert, rapide et sous Apache-2.0":"A small, fast open model under Apache-2.0","Un petit modèle qui mise sur la qualité des données d'entraînement":"A small model focused on training-data quality","Un podcast pour y voir plus clair.":"A podcast to make things clearer.","Un profil publié doit avoir reviewedAt et publishedAt.":"A published profile must have reviewedAt and publishedAt.","Un profil relu doit avoir reviewedAt.":"A reviewed profile must have reviewedAt.","Un projet ne peut figurer qu’une fois dans cette section.":"A project can appear only once in this section.","Un SDK Python léger pour agents multi-étapes":"A lightweight Python SDK for multi-step agents","Un serveur MCP pour piloter un navigateur avec Playwright":"An MCP server for controlling a browser through Playwright","Un snapshot publié ne se modifie pas.":"A published snapshot cannot be changed.","Un sujet, un épisode, une source":"A topic, an episode, a source","Une bibliothèque de skills qui impose une méthode":"A skills library that enforces a method","Une composante absente garde une valeur nulle dans la provenance et son poids est retiré du calcul. Les poids disponibles sont renormalisés. La couverture du score indique leur somme : sans mesure des contributeurs, Momentum utilise 90 % de sa pondération prévue. Le plancher de croissance vaut":"A missing component stays null in the provenance and its weight is removed from the calculation. Available weights are renormalized. Score coverage indicates their sum: without contributor measurements, Momentum uses 90 % of its planned weights. The growth floor is","une fiche figure plusieurs fois.":"a profile appears more than once.","Une interface web auto-hébergée pour modèles locaux et API":"A self-hosted web interface for local models and APIs","Une liste organisée de commandes, skills, hooks et workflows":"An organized list of commands, skills, hooks and workflows","Une mixture d'experts à poids ouverts, pensée pour les agents et le code":"An open-weight mixture of experts for agents and coding","Une mixture d'experts de 235 milliards de paramètres, sous Apache-2.0":"A 235-billion-parameter mixture of experts under Apache-2.0","Une preuve vérifiée ou rejetée doit porter une date et un évaluateur.":"Verified or rejected evidence must include a date and evaluator.","Une semaine figure plusieurs fois dans les éditions.":"A week appears more than once among the editions.","Une traduction publiée doit avoir reviewedAt et publishedAt.":"A published translation must have reviewedAt and publishedAt.","Une traduction relue doit avoir reviewedAt.":"A reviewed translation must have reviewedAt.","URL https requise":"An HTTPS URL is required","Usage d'outils face à un utilisateur simulé.":"Tool use with a simulated user.","Valeur":"Value","Valeur / 100":"Value / 100","Valeur stable sur la période":"Value stable over the period","valeur telle quelle (déjà sur 100)":"raw value (already out of 100)","Value (rapport qualité/prix)":"Value (quality-to-price ratio)","Variation des stars gagnées par les dépôts suivis,":"Change in stars gained by tracked repositories,","Vérifiez les champs requis.":"Check required fields.","Vérifiez les champs.":"Check the fields.","Vidéo":"Video","Vidéo 16:9":"16:9 video","Vidéos":"Videos","Visuel téléchargé.":"Image downloaded.","Vitesse à laquelle le projet gagne de l'attention, en valeur absolue et rapportée à sa taille.":"How quickly the project gains attention, in absolute terms and relative to its size.","Vitesse de génération, échelle logarithmique de 30 à 400 tokens par seconde.":"Generation speed, on a logarithmic scale from 30 to 400 tokens per second.","Vitesse de sortie":"Output speed","Vitesse et prix":"Speed and price","Voir à":"Watch at","Voir la fiche de l’épisode →":"View episode page →","Voir la sélection":"View selection","Voir la semaine":"View week","Voir le":"View the","Voir le chart ↗":"View chart ↗","Voir le dépôt GitHub ↗":"View GitHub repository ↗","Voir le Top":"View Top","Voir sur YouTube":"Watch on YouTube","Voir tous les épisodes":"View all episodes","Voir tous les tops":"View all rankings","Votre adresse email":"Your email address","Vue «":"View \"","Vue par":"View by","Watchlist :":"Watchlist:","Weblog de Simon Willison":"Simon Willison's weblog","Podcasts et classements IA":"Podcasts and AI rankings","{count} épisode":"{count} episode","{count} épisodes":"{count} episodes","{count} épisode trouvé":"{count} episode found","{count} épisodes trouvés":"{count} episodes found","Lire la vidéo : {title}":"Play video: {title}","Lire l'épisode {number} : {title}":"Play episode {number}: {title}","SANDBOX — {title}":"SANDBOX — {title}","Sandbox — {title}":"Sandbox — {title}","Épisode {number} · {title}":"Episode {number} · {title}","{name}, rang {rank}, semaine {week}":"{name}, rank {rank}, week {week}","· filtre {filter}":"· filter {filter}","{count} entrées":"{count} entries","Semaine {week}":"Week {week}","{name} · fiche et historique":"{name} · profile and history","{name} · fiche projet et historique":"{name} · project profile and history","{name} · fiche modèle et historique":"{name} · model profile and history","{title} · Méthode":"{title} · Methodology","{title} · semaine {week}":"{title} · week {week}","Archives · {week}":"Archives · {week}","Aucun résultat pour \"{query}\".":"No results for \"{query}\".","{count} résultat pour \"{query}\"":"{count} result for \"{query}\"","{count} résultats pour \"{query}\"":"{count} results for \"{query}\"","· DIRECT IN THE TOP 10":"· DIRECT IN THE TOP 10","3 MONTHS":"3 MONTHS","30 DAYS":"30 DAYS","30 DAYS MOMENTUM":"30 DAYS MOMENTUM","7 DAYS":"7 DAYS","90 DAYS":"90 DAYS","About":"About","Activity":"Activity","Activity (25 %)":"Activity (25 %)","Agents":"Agents","AI Models Top 20":"AI Models Top 20","AI rankings":"AI rankings","Aider-AI/aider":"Aider-AI/aider","aider.chat":"aider.chat","AIME 2025":"AIME 2025","All":"All","All episodes":"All episodes","ALL TIME":"ALL TIME","All-Hands-AI/OpenHands":"All-Hands-AI/OpenHands","all-hands.dev":"all-hands.dev","Also available in":"Also available in","Anthropic":"Anthropic","anthropics/claude-code":"anthropics/claude-code","anthropics/skills":"anthropics/skills","Apple Podcasts ↗":"Apple Podcasts ↗","Archive":"Archive","ARCHIVE EDITION":"ARCHIVE EDITION","Arena Leaderboard Dataset":"Arena Leaderboard Dataset","Artificial Analysis via OpenRouter":"Artificial Analysis via OpenRouter","astral-sh/uv":"astral-sh/uv","Automation":"Automation","Benchmarks":"Benchmarks","BEST FOR":"BEST FOR","BEST RANK":"BEST RANK","BIGGEST MOVERS":"BIGGEST MOVERS","BIGGEST WEEKLY JUMP":"BIGGEST WEEKLY JUMP","block/goose":"block/goose","Browser":"Browser","browser-use.com":"browser-use.com","browser-use/browser-use":"browser-use/browser-use","BUILDING THE HISTORY":"BUILDING THE HISTORY","Chain-of-Thought Prompting Elicits Reasoning in Large Language Models":"Chain-of-Thought Prompting Elicits Reasoning in Large Language Models","Chapters":"Chapters","Chart":"Chart","CHART INDEX":"CHART INDEX","Charts":"Charts","Charts methodology unavailable:":"Charts methodology unavailable:","Claude Fable 5.1":"Claude Fable 5.1","cline.bot":"cline.bot","cline/cline":"cline/cline","Coding":"Coding","comfy.org":"comfy.org","comfyanonymous/ComfyUI":"comfyanonymous/ComfyUI","COMING NEXT":"COMING NEXT","Context":"Context","context7.com":"context7.com","crewai.com":"crewai.com","crewAIInc/crewAI":"crewAIInc/crewAI","CURRENT RANK":"CURRENT RANK","Current ranking":"Current ranking","Data":"Data","DATA TEMPORARILY UNAVAILABLE":"DATA TEMPORARILY UNAVAILABLE","DeepSeek V3.2":"DeepSeek V3.2","Design":"Design","dify.ai":"dify.ai","Direct Preference Optimization":"Direct Preference Optimization","DISCUSSED IN SANDBOX #":"DISCUSSED IN SANDBOX #","English":"English","EP.":"EP.","Episodes, resources and the latest AI rankings.":"Episodes, resources and the latest AI rankings.","Epoch AI":"Epoch AI","EVERY WEEK LEAVES A TRACE":"EVERY WEEK LEAVES A TRACE","Explore the latest measured AI rankings.":"Explore the latest measured AI rankings.","firecrawl.dev":"firecrawl.dev","firecrawl/firecrawl":"firecrawl/firecrawl","FOLLOW THE ATTENTION":"FOLLOW THE ATTENTION","Forks":"Forks","Gemini 3 Pro":"Gemini 3 Pro","Get SANDBOX CHARTS in your inbox every week.":"Get SANDBOX CHARTS in your inbox every week.","GET THE CHARTS":"GET THE CHARTS","GET THE CHARTS ↗":"GET THE CHARTS ↗","ggml-org/llama.cpp":"ggml-org/llama.cpp","GitHub GraphQL Repository ↗":"GitHub GraphQL Repository ↗","GITHUB PROJECT":"GITHUB PROJECT","GitHub Search API ↗":"GitHub Search API ↗","GitHub star growth":"GitHub star growth","GitHub Top 10":"GitHub Top 10","GITHUB TOP 20":"GITHUB TOP 20","GitHub Trending":"GitHub Trending","github/spec-kit":"github/spec-kit","GLM-4.6":"GLM-4.6","Global":"Global","Google DeepMind":"Google DeepMind","google-gemini/gemini-cli":"google-gemini/gemini-cli","google/adk-python":"google/adk-python","GPQA Diamond":"GPQA Diamond","gpt-oss-120b":"gpt-oss-120b","Growth":"Growth","Growth (55 %)":"Growth (55 %)","GROWTH 7D":"GROWTH 7D","Hacker News":"Hacker News","Hall of #1":"Hall of #1","hesreallyhim/awesome-claude-code":"hesreallyhim/awesome-claude-code","Hugging Face":"Hugging Face","Hugging Face Hub":"Hugging Face Hub","Humanity's Last Exam":"Humanity's Last Exam","ℹ":"ℹ","Image":"Image","Infrastructure":"Infrastructure","Instagram":"Instagram","Install growth":"Install growth","Install reach":"Install reach","Intro":"Intro","Intro to Large Language Models (A. Karpathy)":"Intro to Large Language Models (A. Karpathy)","Kimi K2":"Kimi K2","KNOW WHAT'S":"KNOW WHAT'S","langchain-ai/langgraph":"langchain-ai/langgraph","langgenius/dify":"langgenius/dify","Large language model":"Large language model","Last updated:":"Last updated:","LEADING THIS VIEW":"LEADING THIS VIEW","Let's build GPT: from scratch, in code, spelled out (A. Karpathy)":"Let's build GPT: from scratch, in code, spelled out (A. Karpathy)","LinkedIn ↗":"LinkedIn ↗","LiveCodeBench":"LiveCodeBench","Llama Nemotron Ultra":"Llama Nemotron Ultra","Local AI":"Local AI","Long context":"Long context","LONGEST #1 STREAK":"LONGEST #1 STREAK","Losing momentum this week.":"Losing momentum this week.","Main navigation":"Main navigation","MARKET SIGNALS":"MARKET SIGNALS","Marketing":"Marketing","mastra-ai/mastra":"mastra-ai/mastra","mastra.ai":"mastra.ai","Maths":"Maths","MCP":"MCP","MEASURES, THEN OUR TAKE":"MEASURES, THEN OUR TAKE","Methodology and limits":"Methodology and limits","microsoft/playwright-mcp":"microsoft/playwright-mcp","MiniMax M2":"MiniMax M2","MMMU":"MMMU","MODE 16:9":"MODE 16:9","Model Context Protocol":"Model Context Protocol","modelcontextprotocol.io":"modelcontextprotocol.io","modelcontextprotocol/servers":"modelcontextprotocol/servers","Models":"Models","MODELS TOP 20":"MODELS TOP 20","Momentum":"Momentum","MOST WEEKS IN TOP 20":"MOST EDITIONS IN TOP 20","Movement":"Movement","MRCR":"MRCR","Multimodal":"Multimodal","n8n-io/n8n":"n8n-io/n8n","n8n.io":"n8n.io","Name":"Name","New entries in the charts.":"New entries in the charts.","NEW THIS WEEK":"NEW THIS WEEK","NO ENTRIES":"NO ENTRIES","Not big yet. We're watching them.":"Not big yet. We're watching them.","obra/superpowers":"obra/superpowers","Ollama":"Ollama","ollama.com":"ollama.com","ollama/ollama":"ollama/ollama","OLMo 3":"OLMo 3","ON THE PODCAST":"ON THE PODCAST","Open source":"Open source","Open Source AI Top 10":"Open Source AI Top 10","Open weights":"Open weights","open-webui/open-webui":"open-webui/open-webui","OpenAI":"OpenAI","OpenAI MRCR":"OpenAI MRCR","openai/codex":"openai/codex","openai/openai-agents-python":"openai/openai-agents-python","OpenRouter API":"OpenRouter API","openwebui.com":"openwebui.com","Organization":"Organization","Other":"Other","OUR TAKE":"OUR TAKE","Outro":"Outro","Overall":"Overall","Papers":"Papers","Peak":"Peak","Podcasts":"Podcasts","PREVIOUS RANK":"PREVIOUS RANK","Previous ranking":"Previous ranking","Price":"Price","Productivity":"Productivity","Qwen3 235B":"Qwen3 235B","Rank":"Rank","RANK / MOVEMENT / PROJECT":"RANK / MOVEMENT / PROJECT","Reach":"Reach","Reach (20 %)":"Reach (20 %)","ReAct: Synergizing Reasoning and Acting in Language Models":"ReAct: Synergizing Reasoning and Acting in Language Models","Reasoning":"Reasoning","Repository freshness":"Repository freshness","Research":"Research","Rising":"Rising","RISING 20":"RISING 20","RSS feed":"RSS feed","Run search":"Run search","Sandbox · EP.":"Sandbox · EP.","Sandbox · Podcasts":"Sandbox · Podcasts","Sandbox /":"Sandbox /","SANDBOX CHARTS":"SANDBOX CHARTS","SANDBOX CHARTS ·":"SANDBOX CHARTS ·","SANDBOX CHARTS / EVERY MONDAY":"SANDBOX CHARTS / EVERY MONDAY","SANDBOX CHARTS / MONTHLY EDITION":"SANDBOX CHARTS / MONTHLY EDITION","SANDBOX CHARTS | Rising 20":"SANDBOX CHARTS | Rising 20","SANDBOX CHARTS | The weekly charts of what matters in AI.":"SANDBOX CHARTS | The weekly charts of what matters in AI.","Sandbox home":"Sandbox home","SANDBOX Momentum Score":"SANDBOX Momentum Score","SANDBOX PICK":"SANDBOX PICK","SANDBOX project metadata unavailable:":"SANDBOX project metadata unavailable:","SANDBOX project unavailable:":"SANDBOX project unavailable:","SANDBOX Rising Score":"SANDBOX Rising Score","SANDBOX SCORE":"SANDBOX SCORE","SANDBOX TAKE":"SANDBOX TAKE","Search":"Search","Search podcasts and AI rankings":"Search podcasts and AI rankings","SHARE CHART ↗":"SHARE CHART ↗","SHARE RANKING":"SHARE RANKING","SHARE RANKING ↗":"SHARE RANKING ↗","Sites":"Sites","Skills":"Skills","Skills Top 20":"Skills Top 20","SKILLS TOP 20":"SKILLS TOP 20","Skills.sh API":"Skills.sh API","Skip to content":"Skip to content","Speed":"Speed","Spotify ↗":"Spotify ↗","stars":"stars","Stars":"Stars","stars.":"stars.","Story":"Story","SWE-bench":"SWE-bench","SWE-bench Verified":"SWE-bench Verified","Terminal-Bench":"Terminal-Bench","THE AI CHARTS":"THE AI CHARTS","THE ARCHIVES":"THE ARCHIVES","The archives.":"The archives.","The biggest jumps this week.":"The biggest jumps this week.","THE DISCOVERY CHART":"THE DISCOVERY CHART","THE EDITORIAL SELECTION":"THE EDITORIAL SELECTION","The fastest-moving AI projects on GitHub.":"The fastest-moving AI projects on GitHub.","The models leading AI right now.":"The models leading AI right now.","THE ONE TO BEAT":"THE ONE TO BEAT","THE RECORD BOOK":"THE RECORD BOOK","The skills, agents and MCPs gaining momentum.":"The skills, agents and MCPs gaining momentum.","The weekly charts of what matters in AI.":"The weekly charts of what matters in AI.","THE WEEKLY RANKING":"THE WEEKLY RANKING","THIS MONTH":"THIS MONTH","THIS WEEK":"THIS WEEK","Throughput":"Throughput","TOP 20":"TOP 20","Topics":"Topics","Tweets":"Tweets","UPDATED EVERY MONDAY":"UPDATED EVERY MONDAY","upstash/context7":"upstash/context7","Value":"Value","Version":"Version","Video":"Video","Vision":"Vision","Visual Studio Code":"Visual Studio Code","vllm-project/vllm":"vllm-project/vllm","Voice":"Voice","Watchlist":"Watchlist","Week":"Week","Weekly history":"Weekly history","Weeks in Top 10":"Weeks in Top 10","WEEKS IN TOP 20":"WEEKS IN TOP 20","WHAT IT IS":"WHAT IT IS","What to watch before everyone else.":"What to watch before everyone else.","WHY IT MATTERS":"WHY IT MATTERS","WHY IT'S TRENDING":"WHY IT'S TRENDING","xAI":"xAI","zed-industries/zed":"zed-industries/zed","zed.dev":"zed.dev","τ-bench":"τ-bench","Métadonnées des dépôts reliés aux skills : stars, forks, visibilité et dernière activité déclarée.":"Metadata for repositories linked to skills: stars, forks, visibility and last reported activity.","Identité et compteurs d'installations cumulés du classement Skills.sh.":"Identity and cumulative installation counts from the Skills.sh ranking.","Identifiants, téléchargements sur 30 jours, likes, tags de licence et accès des modèles publics.":"Identifiers, downloads over 30 days, likes, license tags and access to public models.","Scores Arena publiés par catégorie (texte, WebDev, agents, vision et recherche), avec rang et nombre de votes.":"Arena scores published by category (text, WebDev, agents, vision and search), with rank and vote count.","Usage hebdomadaire, débit relatif, fenêtre de contexte et prix par million de tokens.":"Weekly usage, relative throughput, context window and price per million tokens.","Taux de résolution de tickets GitHub réels (SWE-bench Verified).":"Resolution rate for real GitHub issues (SWE-bench Verified).","Problèmes de programmation compétitive récents, pour limiter la contamination.":"Recent competitive programming problems to limit contamination.","Questions de sciences de niveau doctorat, conçues pour résister à la recherche web.":"PhD-level science questions designed to resist web search.","Questions expertes multi-disciplines, scores encore bas pour tous les modèles.":"Expert questions across disciplines, with scores still low for all models.","Évaluations indépendantes, dont les concours de maths (AIME).":"Independent evaluations, including mathematics competitions (AIME).","Tâches d'agents dans un terminal réel.":"Agent tasks in a real terminal.","Agents qui utilisent des outils face à un utilisateur simulé.":"Agents using tools with a simulated user.","Compréhension multimodale (images et texte) de niveau universitaire.":"University-level multimodal understanding (images and text).","Vitesse de génération, latence et prix des API, mesurés de façon homogène.":"API generation speed, latency and pricing, measured consistently.","AI Models":"AI Models","Collecte quotidienne; gel hebdomadaire le lundi après les mises à jour de sources.":"Daily collection; weekly freeze on Monday after source updates.","Classements publics de texte, WebDev, agents, vision et recherche; un score n'est retenu qu'à partir de 1 000 votes, ou 1 000 observations pour la catégorie agents.":"Public rankings for text, WebDev, agents, vision and search; a score is included only after 1 000 votes, or 1 000 observations for the agents category.","Identité des modèles publics, téléchargements sur 30 jours, likes, licence déclarée et accès restreint. Les évaluations de model cards ne sont pas classées tant qu'elles ne sont pas validées indépendamment.":"Public model identity, downloads over 30 days, likes, reported license and restricted access. Model card evaluations are excluded from rankings until independently validated.","Rang d'usage hebdomadaire, rang relatif de débit, contexte maximal et prix publiés par modèle.":"Weekly usage rank, relative throughput rank, maximum context and published model pricing.","Intelligence Index, Coding Index et Agentic Index lus via le point d'accès benchmarks d'OpenRouter.":"Intelligence Index, Coding Index and Agentic Index retrieved through OpenRouter's benchmarks endpoint.","Le score Arena texte et l'Artificial Analysis Intelligence Index sont convertis séparément en percentiles sur les candidats de la semaine, puis moyennés à parts égales. Si une seule source est disponible, son poids est renormalisé.":"The Arena text score and Artificial Analysis Intelligence Index are separately converted to percentiles among the week's candidates, then averaged with equal weights. When only one source is available, its weight is renormalized.","Moyenne des percentiles Arena WebDev et Artificial Analysis Coding Index quand les deux existent; sinon le percentile disponible est utilisé.":"Average of the Arena WebDev and Artificial Analysis Coding Index percentiles when both exist; otherwise the available percentile is used.","Vue désactivée tant qu'aucune évaluation externe dont le jeton de vérification a été contrôlé n'est disponible. Les valeurs auto-déclarées sur les model cards sont exclues.":"View disabled until an external evaluation with a checked verification token is available. Self-reported model card values are excluded.","Vue désactivée tant qu'aucune évaluation mathématique externe vérifiée n'est disponible.":"View disabled until a verified external mathematics evaluation is available.","Percentile du score Arena Agent; complété par l'Agentic Index d'Artificial Analysis quand il existe.":"Percentile of the Arena Agent score, supplemented by the Artificial Analysis Agentic Index when available.","Percentile du score Arena Vision. Les modèles absents de ce classement n'apparaissent pas dans cette vue.":"Percentile of the Arena Vision score. Models absent from this ranking do not appear in this view.","Classement par taille de contexte déclarée par OpenRouter; le contexte ne rentre pas dans le score de qualité générale.":"Ranked by context size reported by OpenRouter; context does not contribute to the overall quality score.","Débit : ordre publié par OpenRouter. Prix : moyenne pondérée de trois parts entrée et une part sortie, par million de tokens; le coût le plus bas obtient le percentile le plus élevé.":"Throughput: order published by OpenRouter. Price: weighted average of three parts input and one part output, per million tokens; the lowest cost receives the highest percentile.","60 % qualité générale et 40 % percentile de prix. C'est un choix éditorial, distinct du score de qualité.":"60 % overall quality and 40 % price percentile. This is an editorial choice, separate from the quality score.","Le pool vient des premières pages publiques de Hugging Face, Arena et OpenRouter. Les sources ne sont reliées que par nom de modèle normalisé et organisation canonique; les collisions restent séparées, sans rapprochement approximatif.":"The candidate pool comes from the first public pages of Hugging Face, Arena and OpenRouter. Sources are linked only by normalized model name and canonical organization; collisions remain separate, without approximate matching.","Un benchmark mesure une tâche, pas votre usage. Aucun score ne remplace un essai sur vos propres données.":"A benchmark measures a task, not your use case. No score replaces a trial on your own data.","Arena mesure des préférences humaines; Artificial Analysis publie un index distinct. Un percentile ne rend pas leurs protocoles identiques.":"Arena measures human preferences; Artificial Analysis publishes a separate index. A percentile does not make their protocols identical.","Les vues Reasoning et Maths restent désactivées jusqu'à validation réelle des scores et de leurs jetons; les résultats auto-déclarés ne sont pas considérés comme vérifiés.":"The Reasoning and Maths views remain disabled until their scores and tokens are actually validated; self-reported results are not treated as verified.","Le débit, le contexte et le prix dépendent du catalogue OpenRouter et de ses fournisseurs; le rang de débit n'est pas un temps de réponse garanti.":"Throughput, context and pricing depend on OpenRouter's catalog and providers; throughput rank is not a guaranteed response time.","Les percentiles changent lorsque le pool de candidats change. Le rang hebdomadaire est conservé, mais un score sur 100 n'est comparable qu'à l'intérieur de son édition.":"Percentiles change when the candidate pool changes. Weekly ranks are preserved, but a score out of 100 is comparable only within its edition.","Le classement ne couvre que les modèles présents dans les sources collectées et les éditions ne sont publiées qu'à partir de vingt candidats évaluables.":"The ranking covers only models present in the collected sources, and editions are published only with at least twenty assessable candidates.","GitHub":"GitHub","Chaque semaine, publication le mardi vers 07 h UTC. Les données sont relevées dans la nuit précédente.":"Every week, published on Tuesday around 07 UTC. Data is collected the previous night.","Toutes les métriques : stars totales, stars gagnées sur 7 jours, forks, contributeurs et commits sur 30 jours.":"All metrics: total stars, stars gained over 7 days, forks, contributors and commits over 30 days.","Stars gagnées sur 7 jours (60 %, échelle logarithmique) et croissance rapportée à la taille du dépôt (40 %). Le second terme permet à un petit projet qui double d'apparaître face à un grand projet qui grossit de 1 %.":"Stars gained over 7 days (60 %, logarithmic scale) and growth relative to repository size (40 %). The second term lets a small project that doubles compete with a large project growing by 1 %.","Commits des 30 derniers jours (60 %) et nombre de contributeurs (40 %), tous deux en échelle logarithmique. Un dépôt qui ne bouge plus ne peut pas monter durablement.":"Commits over the last 30 days (60 %) and contributor count (40 %), both on a logarithmic scale. A repository with no activity cannot keep climbing.","Stars totales (60 %) et forks (40 %), en échelle logarithmique. L'audience installée compte, mais pèse peu : c'est volontaire.":"Total stars (60 %) and forks (40 %), on a logarithmic scale. The established audience counts, but has a small weight by design.","Chaque sous-score est ramené entre 0 et 100 par rapport aux candidats de la semaine (le meilleur vaut 100, le moins bon 0). Seul le rang est donc comparable d'une semaine à l'autre, pas le score.":"Each subscore is scaled between 0 and 100 relative to the week's candidates (the best gets 100, the lowest 0). Only ranks are comparable from one week to the next, not scores.","À score égal, le projet qui a gagné le plus de stars sur 7 jours passe devant.":"When scores tie, the project that gained the most stars over 7 days comes first.","Les stars mesurent l'attention, pas la qualité ni l'usage réel. Un projet très partagé sur les réseaux peut monter vite puis retomber.":"Stars measure attention, not quality or actual usage. A project widely shared on social media can climb quickly and then fall.","Les stars peuvent être gonflées artificiellement. Nous ne filtrons pas encore automatiquement ce bruit ; en cas de pic suspect, l'équipe l'écrit dans OUR TAKE.":"Stars can be artificially inflated. We do not yet filter this noise automatically; when a spike looks suspicious, the team explains it in OUR TAKE.","Le pool de candidats (une quinzaine de dépôts par semaine) est constitué par l'équipe à partir de GitHub Trending, des projets cités dans nos sources et des propositions reçues. Un projet absent du pool ne peut pas entrer.":"The candidate pool (around fifteen repositories per week) is selected by the team from GitHub Trending, projects mentioned in our sources and submitted suggestions. A project outside the pool cannot enter.","La normalisation est relative au pool : ajouter ou retirer un candidat peut déplacer les scores des autres, c'est pourquoi le rang fait foi.":"Normalization is relative to the pool: adding or removing a candidate can shift other scores, which is why rank is the reference.","Open Source AI":"Open Source AI","Chaque semaine, publication le mardi vers 07 h UTC.":"Every week, published on Tuesday around 07 UTC.","Disponibilité des poids, licence et taille du modèle. Un modèle sans poids téléchargeables ne peut pas entrer.":"Weight availability, license and model size. A model without downloadable weights cannot enter.","Coding : résolution de tickets GitHub réels (SWE-bench Verified).":"Coding: resolution of real GitHub issues (SWE-bench Verified).","Coding : problèmes de programmation récents.":"Coding: recent programming problems.","Reasoning : questions de sciences de niveau doctorat.":"Reasoning: PhD-level science questions.","Reasoning : questions expertes.":"Reasoning: expert questions.","Maths : concours AIME.":"Maths: AIME competitions.","Agents : tâches dans un terminal.":"Agents: tasks in a terminal.","Agents : usage d'outils face à un utilisateur simulé.":"Agents: tool use with a simulated user.","Multimodal : compréhension d'images et de texte.":"Multimodal: understanding images and text.","Long context : retrouver l'information dans un long contexte.":"Long context: finding information in a long context.","Fenêtre de contexte, vitesse de sortie et prix des API d'hébergement.":"Context window, output speed and hosting API pricing.","Critère d'entrée : les poids doivent être téléchargeables. Nous ne jugeons pas la licence : elle est affichée sur chaque fiche, car « open weights » ne veut pas dire « libre ».":"Entry requirement: weights must be downloadable. We do not judge the license; it is displayed on each profile because open weights does not mean unrestricted use.","Identique à l'AI Models Top 10 : coding 25 %, reasoning 22 %, agents 20 %, maths 13 %, multimodal 10 %, long context 10 %.":"Identical to the AI Models Top 10: coding 25 %, reasoning 22 %, agents 20 %, maths 13 %, multimodal 10 %, long context 10 %.","Pour un modèle ouvert, la vitesse et le prix sont ceux des fournisseurs qui l'hébergent (médiane). Auto-hébergé, votre coût dépendra de votre matériel.":"For an open model, speed and price are those of its hosting providers (median). If you self-host, your cost depends on your hardware.","Les licences varient beaucoup : Apache-2.0, MIT, licences maison avec restrictions. Lisez la licence avant de bâtir un produit dessus.":"Licenses vary widely: Apache-2.0, MIT, and custom licenses with restrictions. Read the license before building a product on it.","Les scores de certains modèles viennent de l'éditeur tant qu'aucune évaluation indépendante n'existe. Ils sont signalés comme tels sur la fiche.":"Some model scores come from the publisher until an independent evaluation exists. They are labeled as such on the profile.","Les modèles texte seul n'ont pas de score multimodal ; ils ne sont pas pénalisés dans le Global.":"Text-only models have no multimodal score; they are not penalized in the Global view.","Le coût réel d'un modèle auto-hébergé n'est pas mesuré ici.":"The actual cost of a self-hosted model is not measured here.","Relevés quotidiens vers 02 h 30 UTC; gel hebdomadaire le lundi après la collecte.":"Daily observations around 02:30 UTC; weekly freeze on Monday after collection.","Identité du skill et compteur cumulatif d'installations; les variations sont calculées à partir de nos relevés quotidiens.":"Skill identity and cumulative installation count; changes are calculated from our daily observations.","Vérifie le dépôt public associé et fournit ses stars, forks et date de dernier push.":"Verifies the linked public repository and provides its stars, forks and last push date.","Variation du compteur d'installations entre le relevé courant et le relevé le plus proche de J−7, normalisée en percentile parmi les candidats de la semaine.":"Change in installation count between the current observation and the nearest observation to D−7, normalized as a percentile among the week's candidates.","Stars gagnées entre les deux relevés espacés d'environ sept jours, normalisées en percentile.":"Stars gained between observations approximately seven days apart, normalized as a percentile.","Décroissance exponentielle à partir de la date du dernier push, avec une constante de temps de 30 jours; le sous-score vaut environ 37 % après 30 jours sans activité.":"Exponential decay from the last push date, with a time constant of 30 days; the subscore is about 37 % after 30 days without activity.","Le compteur total d'installations Skills.sh est affiché en percentile comme vue secondaire; il ne modifie pas le score principal.":"The total Skills.sh installation count is shown as a percentile in a secondary view; it does not affect the main score.","Les deux premières pages de Skills.sh en vues all-time et trending alimentent le pool. Les doublons, dépôts privés, archivés, forks et dépôts désactivés sont écartés. Un dépôt peut aussi apparaître dans le GitHub Top 20.":"The first two Skills.sh pages in the all-time and trending views supply the pool. Duplicates, private repositories, archived repositories, forks and disabled repositories are excluded. A repository may also appear in the GitHub Top 20.","Les sous-scores sont des percentiles recalculés parmi les candidats disposant des relevés complets. Les égalités reçoivent leur rang moyen. Une édition n'est gelée qu'avec au moins vingt candidats et sept jours d'historique.":"Subscores are percentiles recalculated among candidates with complete observations. Ties receive their average rank. An edition is frozen only with at least twenty candidates and seven days of history.","Skills.sh publie un compteur cumulatif : l'évolution hebdomadaire est calculée par nos relevés et ne peut pas être reconstruite avant leur démarrage.":"Skills.sh publishes a cumulative count: weekly changes are calculated from our observations and cannot be reconstructed before collection begins.","La fraîcheur du dépôt est un indicateur de mise à jour, pas une preuve de qualité ou de bon fonctionnement.":"Repository freshness indicates updates, not proof of quality or correct operation.","Les rangs sont relatifs au pool de Skills.sh observé et aux dépôts GitHub reliés; ils ne couvrent pas tous les skills disponibles sur Internet.":"Ranks are relative to the observed Skills.sh pool and linked GitHub repositories; they do not cover every skill available on the Internet.","Une édition ne paraît pas tant que vingt candidats n'ont pas les relevés requis.":"An edition is not published until twenty candidates have the required observations.","Semaine 41 : les mouvements des quatre classements, sans le blabla":"Week 41: movements across the four rankings, without the filler","Claude Code prend la tête du GitHub Top 10, adk-python bondit de quatre places, Gemini 3 Pro revient sur le podium des modèles.":"Claude Code leads the GitHub Top 10, adk-python jumps four places, and Gemini 3 Pro returns to the models podium.","Chaque mardi, les quatre classements sont recalculés à partir des nouvelles données. Voici ce qui a bougé en semaine 41, dans l'ordre où on en parle dans l'[[episode:42]].":"Every Tuesday, the four rankings are recalculated from fresh data. Here is what changed in week 41, in the order discussed in [[episode:42]].","Chaque mardi, les quatre classements sont recalculés à partir des nouvelles données. Voici ce qui a bougé en semaine 41, dans l'ordre où on en parle dans l'":"Every Tuesday, the four rankings are recalculated from fresh data. Here is what changed in week 41, in the order discussed in","**↑ 1** [[entity:claude-code]] passe #1 avec 14 406 stars gagnées en sept jours.":"**↑ 1** [[entity:claude-code]] moves to #1 with 14 406 stars gained in seven days.","passe #1 avec 14 406 stars gagnées en sept jours.":"moves to #1 with 14 406 stars gained in seven days.","**↓ 2** [[entity:mastra]] recule à la 5ᵉ place, malgré 3 294 nouvelles stars.":"**↓ 2** [[entity:mastra]] falls to 5th place despite 3 294 new stars.","recule à la 5ᵉ place, malgré 3 294 nouvelles stars.":"falls to 5th place despite 3 294 new stars.","**RE** [[entity:dify]] revient dans le Top 10, à la 9ᵉ place.":"**RE** [[entity:dify]] returns to the Top 10 in 9th place.","RE":"RE","revient dans le Top 10, à la 9ᵉ place.":"returns to the Top 10 in 9th place.","**OUT** [[entity:browser-use]] sort du Top 10 (11ᵉ) après son pic de l'été.":"**OUT** [[entity:browser-use]] leaves the Top 10 (11th) after its summer peak.","OUT":"OUT","sort du Top 10 (11ᵉ) après son pic de l'été.":"leaves the Top 10 (11th) after its summer peak.","GitHub Skills Top 10":"GitHub Skills Top 10","**=** [[entity:anthropic-skills]] garde la première place pour la troisième semaine.":"**=** [[entity:anthropic-skills]] keeps first place for the third week.","garde la première place pour la troisième semaine.":"keeps first place for the third week.","**↑ 4** [[entity:adk-python]] saute de la 6ᵉ à la 2ᵉ place.":"**↑ 4** [[entity:adk-python]] jumps from 6th to 2nd place.","saute de la 6ᵉ à la 2ᵉ place.":"jumps from 6th to 2nd place.","[[entity:superpowers]] signe le plus gros gain de stars de la semaine (9 455), mais reste 3ᵉ : le classement pèse aussi l'activité et l'audience.":"[[entity:superpowers]] records the week's biggest star gain (9 455), but remains 3rd: the ranking also weighs activity and audience.","signe le plus gros gain de stars de la semaine (9 455), mais reste 3ᵉ : le classement pèse aussi l'activité et l'audience.":"records the week's biggest star gain (9 455), but remains 3rd: the ranking also weighs activity and audience.","**↓ 2** [[entity:browser-use]] tombe à la 4ᵉ place.":"**↓ 2** [[entity:browser-use]] falls to 4th place.","tombe à la 4ᵉ place.":"falls to 4th place.","AI Models Top 10":"AI Models Top 10","**=** [[entity:claude-fable-5-1]] reste #1 pour la cinquième semaine.":"**=** [[entity:claude-fable-5-1]] remains #1 for the fifth week.","reste #1 pour la cinquième semaine.":"remains #1 for the fifth week.","**↑ 2** [[entity:gemini-3-pro]] revient en 2ᵉ position après la révision de ses résultats d'agent.":"**↑ 2** [[entity:gemini-3-pro]] returns to 2nd place after its agent results were revised.","revient en 2ᵉ position après la révision de ses résultats d'agent.":"returns to 2nd place after its agent results were revised.","**↓ 1** [[entity:claude-opus-5-5]] et [[entity:gpt-5-2]] cèdent chacun une place.":"**↓ 1** [[entity:claude-opus-5-5]] and [[entity:gpt-5-2]] each lose one place.","et":"and","cèdent chacun une place.":"each lose one place.","**RE** [[entity:gpt-oss-120b]] retrouve le Top 10 (10ᵉ), et **OUT** [[entity:glm-4-6]] en sort (11ᵉ).":"**RE** [[entity:gpt-oss-120b]] returns to the Top 10 (10th), and **OUT** [[entity:glm-4-6]] leaves it (11th).","retrouve le Top 10 (10ᵉ), et":"returns to the Top 10 (10th), and","en sort (11ᵉ).":"leaves it (11th).","**↑ 3** [[entity:gpt-oss-120b]] grimpe à la 3ᵉ place.":"**↑ 3** [[entity:gpt-oss-120b]] climbs to 3rd place.","grimpe à la 3ᵉ place.":"climbs to 3rd place.","**NEW** [[entity:nemotron-ultra]] entre à la 9ᵉ place.":"**NEW** [[entity:nemotron-ultra]] enters in 9th place.","NEW":"NEW","entre à la 9ᵉ place.":"enters in 9th place.","**OUT** [[entity:llama-4-maverick]] quitte le Top 10 (11ᵉ).":"**OUT** [[entity:llama-4-maverick]] leaves the Top 10 (11th).","quitte le Top 10 (11ᵉ).":"leaves the Top 10 (11th).","**=** [[entity:deepseek-v3-2]] garde la tête pour la treizième semaine.":"**=** [[entity:deepseek-v3-2]] keeps the lead for the thirteenth week.","garde la tête pour la treizième semaine.":"keeps the lead for the thirteenth week.","Un rang compare des projets entre eux la même semaine. Le score, lui, est relatif aux candidats du moment : seul le rang est comparable d'une semaine à l'autre. Tout est expliqué dans [[chart:github]].":"A rank compares projects within the same week. Scores are relative to the current candidates; only ranks are comparable from one week to the next. The method is explained in [[chart:github]].","Un rang compare des projets entre eux la même semaine. Le score, lui, est relatif aux candidats du moment : seul le rang est comparable d'une semaine à l'autre. Tout est expliqué dans":"A rank compares projects within the same week. Scores are relative to the current candidates; only ranks are comparable from one week to the next. The method is explained in","Comment lire ces chiffres":"How to read these figures","gpt-oss-120b grimpe à la 3ᵉ place des modèles ouverts, NVIDIA entre dans le Top 10":"gpt-oss-120b climbs to 3rd place among open models, NVIDIA enters the Top 10","Trois places gagnées pour le modèle à poids ouverts d'OpenAI. Une nouvelle entrée, une sortie : le Top 10 open source de la semaine 41.":"Three places gained for OpenAI's open-weight model. One entry and one exit: the open source Top 10 for week 41.","Dans l'[[chart:open-source-ai]], [[entity:gpt-oss-120b]] passe de la 6ᵉ à la 3ᵉ place. [[entity:deepseek-v3-2]] garde la tête pour la treizième semaine consécutive, devant [[entity:kimi-k2]].":"In [[chart:open-source-ai]], [[entity:gpt-oss-120b]] moves from 6th to 3rd place. [[entity:deepseek-v3-2]] keeps the lead for the thirteenth consecutive week, ahead of [[entity:kimi-k2]].","Dans l'":"In","passe de la 6ᵉ à la 3ᵉ place.":"moves from 6th to 3rd place.","garde la tête pour la treizième semaine consécutive, devant":"keeps the lead for the thirteenth consecutive week, ahead of","places pour gpt-oss-120b":"places for gpt-oss-120b","nouvelle position":"new position","Llama Nemotron Ultra, 9ᵉ":"Llama Nemotron Ultra, 9th","Llama 4 Maverick, 11ᵉ":"Llama 4 Maverick, 11th","Ce qui a bougé":"What changed","Les scores d'agent de gpt-oss-120b ont été révisés à la hausse (SWE-bench Verified de 58 à 68, Terminal-Bench de 32 à 45). Combiné à sa vitesse, la plus élevée du Top 10 des poids ouverts, cela suffit à lui faire dépasser [[entity:glm-4-6]], [[entity:minimax-m2]] et [[entity:qwen3-235b]].":"gpt-oss-120b's agent scores were revised upwards (SWE-bench Verified from 58 to 68, Terminal-Bench from 32 to 45). Combined with its speed, the highest in the open-weight Top 10, this is enough to put it ahead of [[entity:glm-4-6]], [[entity:minimax-m2]] and [[entity:qwen3-235b]].","Les scores d'agent de gpt-oss-120b ont été révisés à la hausse (SWE-bench Verified de 58 à 68, Terminal-Bench de 32 à 45). Combiné à sa vitesse, la plus élevée du Top 10 des poids ouverts, cela suffit à lui faire dépasser":"gpt-oss-120b's agent scores were revised upwards (SWE-bench Verified from 58 to 68, Terminal-Bench from 32 to 45). Combined with its speed, the highest in the open-weight Top 10, this is enough to put it ahead of","[[entity:nemotron-ultra]], un dérivé de Llama affiné par NVIDIA, entre à la 9ᵉ place. Il pousse [[entity:llama-4-maverick]] hors du Top 10 : le multimodal et le très long contexte de ce dernier ne suffisent plus à compenser ses scores d'agent.":"[[entity:nemotron-ultra]], a Llama derivative fine-tuned by NVIDIA, enters in 9th place. It pushes [[entity:llama-4-maverick]] out of the Top 10: the latter's multimodal capabilities and very long context no longer compensate for its agent scores.",", un dérivé de Llama affiné par NVIDIA, entre à la 9ᵉ place. Il pousse":", a Llama derivative fine-tuned by NVIDIA, enters in 9th place. It pushes","hors du Top 10 : le multimodal et le très long contexte de ce dernier ne suffisent plus à compenser ses scores d'agent.":"out of the Top 10: its multimodal capabilities and very long context no longer compensate for its agent scores.","Les licences diffèrent d'un modèle à l'autre : Apache-2.0 pour gpt-oss-120b, licences maison avec conditions d'usage pour d'autres. Elle est affichée sur chaque fiche, lisez-la avant de bâtir un produit dessus.":"Licenses differ between models: Apache-2.0 for gpt-oss-120b, custom licenses with usage conditions for others. The license is displayed on each profile; read it before building a product on the model.","Poids ouverts, pas forcément libres":"Open weights do not always mean unrestricted use","gpt-oss-120b sur Hugging Face":"gpt-oss-120b on Hugging Face","Claude Code prend la première place du GitHub Top 10":"Claude Code takes first place in the GitHub Top 10","Seize semaines dans le classement, un seul passage en tête : celui-ci. Ce que disent les chiffres, et ce qu'ils ne disent pas.":"Sixteen weeks in the ranking and just one stint at the top: this one. What the figures say, and what they leave out.","[[entity:claude-code]] passe #1 du [[chart:github]] cette semaine, en gagnant une place sur [[entity:vllm]], qui tenait la tête depuis trois semaines. C'est la première fois que le projet atteint la première place.":"[[entity:claude-code]] moves to #1 in [[chart:github]] this week, gaining one place over [[entity:vllm]], which had led for three weeks. It is the project's first time in first place.","passe #1 du":"moves to #1 in","cette semaine, en gagnant une place sur":"this week, gaining one place over",", qui tenait la tête depuis trois semaines. C'est la première fois que le projet atteint la première place.":", which had led for three weeks. It is the project's first time in first place.","stars gagnées en 7 jours":"stars gained in 7 days","stars au total":"total stars","croissance hebdomadaire":"weekly growth","semaines dans le Top 10":"weeks in the Top 10","Pourquoi il passe devant":"Why it takes the lead","Le classement pèse la dynamique avant la taille : 55 % de la note vient de la croissance. Sur ce critère, Claude Code obtient le score maximum du pool, parce qu'il cumule le plus fort gain de stars et la plus forte croissance relative.":"The ranking weights momentum ahead of size: 55 % of the score comes from growth. Claude Code gets the maximum score in the pool on this criterion because it combines the biggest star gain with the strongest relative growth.","En revanche, il est dernier ou presque sur l'activité du dépôt (164 commits sur 30 jours, contre 1 226 pour vLLM). C'est un rappel utile : le dépôt public d'un produit n'est pas son code source, et un score d'activité faible ne dit rien de la qualité de l'outil.":"However, it is near the bottom for repository activity (164 commits over 30 days, compared with 1 226 for vLLM). A useful reminder: a product's public repository is not its source code, and low activity says nothing about the tool's quality.","Une star mesure l'attention. Elle ne dit ni si l'outil fonctionne pour vous, ni s'il continuera de monter. On a vu cet été [[entity:browser-use]] passer en tête sur un pic de partages, puis retomber.":"A star measures attention. It does not tell you whether the tool works for you or whether it will keep climbing. This summer, [[entity:browser-use]] took the lead after a spike in shares, then fell back.","Une star mesure l'attention. Elle ne dit ni si l'outil fonctionne pour vous, ni s'il continuera de monter. On a vu cet été":"A star measures attention. It does not tell you whether the tool works for you or whether it will keep climbing. This summer, we saw","passer en tête sur un pic de partages, puis retomber.":"take the lead after a spike in shares, then fall back.","Ce que les stars ne disent pas":"What stars do not tell you","On en parle dans l'[[episode:42]], avec les trois autres classements de la semaine. Notre avis est sur la page du projet, séparé des chiffres.":"We discuss it in [[episode:42]], alongside the week's three other rankings. Our take is on the project's profile, separate from the figures.","On en parle dans l'":"We discuss it in",", avec les trois autres classements de la semaine. Notre avis est sur la page du projet, séparé des chiffres.":", alongside the week's three other rankings. Our take is on the project's profile, separate from the figures.","Dépôt anthropics/claude-code":"anthropics/claude-code repository","Ce que SWE-bench ne dit pas, et pourquoi Gemini 3 Pro gagne deux places":"What SWE-bench leaves out, and why Gemini 3 Pro gains two places","Une révision de scores d'agent suffit à déplacer un modèle de deux rangs. Lecture critère par critère du plus gros mouvement de la semaine.":"Revised agent scores are enough to move a model up two ranks. A criterion-by-criterion look at the week's biggest movement.","[[entity:gemini-3-pro]] passe de la 4ᵉ à la 2ᵉ place de l'[[chart:ai-models]]. Rien n'a été publié par Google cette semaine : ce sont les résultats qui ont été révisés. C'est l'occasion de regarder ce qu'un benchmark mesure vraiment.":"[[entity:gemini-3-pro]] moves from 4th to 2nd place in [[chart:ai-models]]. Google released nothing this week; the results were revised. This is a chance to look at what a benchmark actually measures.","passe de la 4ᵉ à la 2ᵉ place de l'":"moves from 4th to 2nd place in",". Rien n'a été publié par Google cette semaine : ce sont les résultats qui ont été révisés. C'est l'occasion de regarder ce qu'un benchmark mesure vraiment.":". Google released nothing this week; the results were revised. This is a chance to look at what a benchmark actually measures.","Ce qui a changé":"What changed","points sur SWE-bench Verified (76 → 81)":"points on SWE-bench Verified (76 → 81)","points sur Terminal-Bench (49 → 57)":"points on Terminal-Bench (49 → 57)","points sur τ-bench (78 → 85)":"points on τ-bench (78 → 85)","sur le sous-score Agents (73,6 → 83,0)":"on the Agents subscore (73,6 → 83,0)","Ce gain sur l'axe Agents suffit à faire passer la note globale de 81,5 à 84,1, et à dépasser [[entity:claude-opus-5-5]] et [[entity:gpt-5-2]]. [[entity:claude-fable-5-1]] reste devant, avec 85,1.":"This gain on the Agents axis raises the overall score from 81,5 to 84,1, enough to move ahead of [[entity:claude-opus-5-5]] and [[entity:gpt-5-2]]. [[entity:claude-fable-5-1]] remains ahead at 85,1.","Ce gain sur l'axe Agents suffit à faire passer la note globale de 81,5 à 84,1, et à dépasser":"This gain on the Agents axis raises the overall score from 81,5 to 84,1, enough to move ahead of","reste devant, avec 85,1.":"remains ahead at 85,1.","Ce que SWE-bench mesure":"What SWE-bench measures","SWE-bench Verified demande à un modèle de corriger de vrais tickets issus de dépôts publics, puis vérifie que les tests passent. C'est un bon signal sur la capacité à lire du code existant. Mais le score dépend autant du modèle que de l'échafaudage autour : les outils fournis, le nombre d'essais, la façon de présenter le dépôt.":"SWE-bench Verified asks a model to fix real issues from public repositories, then checks that the tests pass. It is a useful signal of the ability to read existing code. But the score depends as much on the surrounding scaffold as on the model: the available tools, the number of attempts and how the repository is presented.","**Il mesure des tickets Python** : peu de front-end, peu de langages typés, peu de monorepos.":"**It measures Python issues**: limited front-end work, typed languages and monorepos.","Il mesure des tickets Python":"It measures Python issues",": peu de front-end, peu de langages typés, peu de monorepos.":": limited front-end work, typed languages and monorepos.","**Il ne dit rien du coût** : deux modèles à 80 % ne consomment pas la même chose pour y arriver.":"**It says nothing about cost**: two models scoring 80 % do not use the same resources to get there.","Il ne dit rien du coût":"It says nothing about cost",": deux modèles à 80 % ne consomment pas la même chose pour y arriver.":": two models scoring 80 % do not use the same resources to get there.","**Il vieillit** : les tickets finissent par fuiter dans les données d'entraînement.":"**It ages**: issues eventually leak into training data.","Il vieillit":"It ages",": les tickets finissent par fuiter dans les données d'entraînement.":": issues eventually leak into training data.","Aucun score ne suffit. Le sous-score Coding combine SWE-bench Verified et LiveCodeBench, l'axe Agents combine Terminal-Bench et τ-bench. Toutes les pondérations sont publiées dans la méthodologie du classement.":"No single score is enough. The Coding subscore combines SWE-bench Verified and LiveCodeBench; the Agents axis combines Terminal-Bench and τ-bench. All weights are published in the ranking methodology.","Pourquoi on en utilise plusieurs":"Why we use several benchmarks","Pourquoi tout le monde parle de spec-driven development":"Why everyone is talking about spec-driven development","Écrire la spécification avant de demander du code à un agent : méthode sérieuse ou nouvelle mode ? Ce qui est vraiment nouveau, et ce qui ne l'est pas.":"Writing a specification before asking an agent for code: a serious method or a new trend? What is actually new, and what is not.","Il y a un an, tout le monde parlait de « vibe coding » : on décrit, l'agent écrit, on regarde si ça marche. Aujourd'hui, les projets qui montent dans le [[chart:skills]] disent presque tous la même chose : écrivez d'abord une spec.":"A year ago, everyone talked about vibe coding: describe it, let the agent write it, then see if it works. Today, projects climbing [[chart:skills]] nearly all say the same thing: write a spec first.","Il y a un an, tout le monde parlait de « vibe coding » : on décrit, l'agent écrit, on regarde si ça marche. Aujourd'hui, les projets qui montent dans le":"A year ago, everyone talked about vibe coding: describe it, let the agent write it, then see if it works. Today, projects climbing","disent presque tous la même chose : écrivez d'abord une spec.":"nearly all say the same thing: write a spec first.","L'idée, sans le jargon":"The idea, without the jargon","Au lieu de demander une fonctionnalité en une phrase, on écrit un document qui dit ce qu'elle doit faire, ce qu'elle ne doit pas faire et comment on saura qu'elle marche. L'agent produit ensuite un plan, des tâches, puis le code. [[entity:spec-kit]] formalise cette chaîne, [[entity:superpowers]] en impose une variante sous forme de skills.":"Instead of requesting a feature in one sentence, write a document explaining what it should do, what it should avoid and how to know it works. The agent then produces a plan, tasks and code. [[entity:spec-kit]] formalizes this sequence; [[entity:superpowers]] enforces a variant through skills.","Au lieu de demander une fonctionnalité en une phrase, on écrit un document qui dit ce qu'elle doit faire, ce qu'elle ne doit pas faire et comment on saura qu'elle marche. L'agent produit ensuite un plan, des tâches, puis le code.":"Instead of requesting a feature in one sentence, write a document explaining what it should do, what it should avoid and how to know it works. The agent then produces a plan, tasks and code.","formalise cette chaîne,":"formalizes this sequence;","en impose une variante sous forme de skills.":"enforces a variant through skills.","Ce qui est vraiment nouveau":"What is actually new","La spec devient **le document de référence** : le code peut être régénéré, la spec reste.":"The spec becomes **the reference document**: code can be regenerated; the spec remains.","La spec devient":"The spec becomes","le document de référence":"the reference document",": le code peut être régénéré, la spec reste.":": code can be regenerated; the spec remains.","Elle sert **d'interface avec l'agent** : on relit un texte de deux pages plutôt que mille lignes de diff.":"It serves as **the interface with the agent**: review two pages of text instead of a thousand lines of diff.","Elle sert":"It serves as","d'interface avec l'agent":"the interface with the agent",": on relit un texte de deux pages plutôt que mille lignes de diff.":": review two pages of text instead of a thousand lines of diff.","Les critères d'acceptation sont écrits **avant**, ce qui rend les tests plus difficiles à tricher.":"Acceptance criteria are written **in advance**, making tests harder to game.","Les critères d'acceptation sont écrits":"Acceptance criteria are written","avant":"in advance",", ce qui rend les tests plus difficiles à tricher.":", making tests harder to game.","Ce qui ne l'est pas":"What is not new","Écrire une spécification avant de coder, c'est le cahier des charges des années 90, avec ses défauts connus : on découvre les vrais problèmes en codant. La différence, c'est que l'itération coûte désormais quelques minutes d'agent plutôt que quelques semaines d'équipe.":"Writing a specification before coding recalls requirements documents from the 90s, with their familiar flaws: the real problems emerge during implementation. The difference is that iteration now costs a few minutes of agent time instead of a few weeks of team time.","Une spec trop détaillée, c'est du code mal écrit dans un fichier Markdown.":"An overly detailed spec is poorly written code in a Markdown file.","Équipe Sandbox, contenu simulé":"Sandbox team, simulated content","Sur un petit changement, la cérémonie coûte plus cher que le code. Sur un gros chantier flou, la spec peut donner une fausse impression de maîtrise. Notre règle du moment : spec courte, critères d'acceptation précis, et on jette ce qui ne sert pas.":"For a small change, the ceremony costs more than the code. For a large, unclear project, a spec can create a false sense of control. Our current rule: a short spec, precise acceptance criteria, and discard anything that does not help.","Là où ça coince":"Where it gets difficult","On a passé une bonne partie de l'[[episode:40]] sur le sujet, et on y revient dans l'[[episode:42]]. Les deux projets cités ont leur fiche, avec leur historique de classement.":"We spent much of [[episode:40]] on this topic and return to it in [[episode:42]]. Both projects have profiles with their ranking history.","On a passé une bonne partie de l'":"We spent much of","sur le sujet, et on y revient dans l'":"on this topic and return to it in",". Les deux projets cités ont leur fiche, avec leur historique de classement.":". Both projects have profiles with their ranking history.","Trois agents de code, un même dépôt, une même consigne : le protocole":"Three coding agents, one repository, one instruction: the protocol","On ne publie pas encore de résultats. On publie d'abord la façon dont on va les mesurer, pour que personne ne puisse nous accuser d'avoir truqué le match.":"We are not publishing results yet. First, we publish how we will measure them, so nobody can accuse us of rigging the match.","L'idée est simple : prendre un vieux dépôt avec une dette réelle, donner la même consigne à [[entity:claude-code]], [[entity:codex]] et [[entity:gemini-cli]], et comparer ce qui sort. Les comparaisons d'agents sont souvent bâclées. On préfère annoncer les règles avant.":"The idea is simple: take an old repository with real technical debt, give the same instruction to [[entity:claude-code]], [[entity:codex]] and [[entity:gemini-cli]], and compare the output. Agent comparisons are often rushed. We prefer to announce the rules first.","L'idée est simple : prendre un vieux dépôt avec une dette réelle, donner la même consigne à":"The idea is simple: take an old repository with real technical debt, give the same instruction to",", et comparer ce qui sort. Les comparaisons d'agents sont souvent bâclées. On préfère annoncer les règles avant.":", and compare the output. Agent comparisons are often rushed. We prefer to announce the rules first.","Le protocole":"The protocol","Un dépôt unique, choisi avant de lancer quoi que ce soit, avec une suite de tests qui passe au départ.":"One repository, chosen before anything runs, with a test suite that passes at the start.","Une consigne écrite une fois : migrer la couche d'accès aux données vers une bibliothèque plus récente, sans changer le comportement visible.":"One written instruction: migrate the data access layer to a newer library without changing visible behavior.","Trois exécutions par agent, mêmes outils autorisés, même limite de temps, aucune relance humaine.":"Three runs per agent, the same permitted tools, the same time limit and no human follow-up.","Un seul critère de succès : la suite de tests existante passe **et** une relecture à l'aveugle par deux personnes qui ne savent pas quel agent a produit quoi.":"One success criterion: the existing test suite passes **and** two reviewers assess the result without knowing which agent produced it.","Un seul critère de succès : la suite de tests existante passe":"One success criterion: the existing test suite passes","une relecture à l'aveugle par deux personnes qui ne savent pas quel agent a produit quoi.":"two reviewers assess the result without knowing which agent produced it.","Ce qu'on mesurera":"What we will measure","Taux de réussite sur les trois essais.":"Success rate over the three attempts.","Temps écoulé et coût en tokens.":"Elapsed time and token cost.","Nombre de fichiers modifiés hors périmètre.":"Number of files changed outside the requested scope.","Tests supprimés ou affaiblis pour « faire passer » (on en tient un décompte à part).":"Tests deleted or weakened to make the suite pass (tracked separately).","Aucun résultat n'est publié sur cette page tant que les neuf exécutions ne sont pas terminées et relues. Nous ne ferons pas de classement d'agents à partir d'un seul essai.":"No results are published on this page until all nine runs are complete and reviewed. We will not rank agents from a single attempt.","Expérience en cours":"Experiment in progress","Les résultats seront publiés dans un article dédié et commentés dans un épisode. Si le protocole vous paraît biaisé, c'est le moment de nous le dire.":"Results will be published in a dedicated article and discussed in an episode. If the protocol seems biased, now is the time to tell us.","Les benchmarks d'agents sont le nouveau « ça marche sur ma machine »":"Agent benchmarks are the new 'it works on my machine'","Un score d'agent dépend autant de l'échafaudage que du modèle. Tant qu'on ne publie pas l'échafaudage, on compare des cuisines, pas des cuisiniers.":"An agent score depends as much on its scaffold as on the model. Until the scaffold is published, we are comparing kitchens rather than cooks.","Quand un éditeur annonce 80 % sur un benchmark d'agent, personne ne demande dans quelle configuration. Le modèle ? Oui. Les outils fournis ? Le nombre d'essais ? Le prompt système ? La température ? Ce sont pourtant ces détails qui font passer un score de 60 à 80.":"When a publisher announces 80 % on an agent benchmark, nobody asks about the configuration. The model? Yes. But the tools? Number of attempts? System prompt? Temperature? These details are what take a score from 60 to 80.","Un benchmark d'agent sans son échafaudage, c'est une recette sans le four.":"An agent benchmark without its scaffold is a recipe without an oven.","Ce qu'on voudrait voir":"What we would like to see","Le harnais d'évaluation publié, pas seulement le score.":"The evaluation harness published alongside the score.","Le coût de chaque point : tokens, appels d'outils, temps.":"The cost of each point: tokens, tool calls and time.","La variance sur plusieurs essais : un 80 % moyen avec plus ou moins 15 points n'est pas un 80 %.":"Variance across several attempts: an average of 80 % with a spread of plus or minus 15 points is not simply 80 %.","La mention de qui a mesuré : l'éditeur, un tiers, ou nous.":"Who measured it: the publisher, a third party or us.","C'est une des raisons pour lesquelles notre [[chart:ai-models]] combine plusieurs sources et affiche les limites de chacune. Cela ne règle pas le problème, cela le rend visible. Pour un cas concret, voir [[story:ce-que-swe-bench-ne-dit-pas]].":"This is one reason our [[chart:ai-models]] combines several sources and displays each source's limitations. It does not solve the problem; it makes it visible. For a concrete example, see [[story:ce-que-swe-bench-ne-dit-pas]].","C'est une des raisons pour lesquelles notre":"This is one reason our","combine plusieurs sources et affiche les limites de chacune. Cela ne règle pas le problème, cela le rend visible. Pour un cas concret, voir":"combines several sources and displays each source's limitations. It does not solve the problem; it makes it visible. For a concrete example, see","Cette page est une opinion de l'équipe. Les données des classements, elles, sont séparées et sourcées.":"This page expresses the team's opinion. Ranking data is kept separate and sourced.","C'est un avis":"An opinion","Lire un classement de modèles sans se faire avoir":"Reading a model ranking without being misled","Cinq réflexes pour comprendre un leaderboard, le nôtre compris : quel critère, qui a mesuré, quel écart, quelle date, quel prix.":"Five habits for understanding a leaderboard, including ours: the criterion, who measured it, the gap, the date and the price.","On nous demande souvent quel est « le meilleur modèle ». La réponse honnête, c'est : meilleur pour quoi, pour qui, à quel prix. Voici comment lire un classement, le nôtre comme les autres.":"We are often asked which model is best. The honest answer is: best for what, for whom and at what price? Here is how to read a ranking, ours or anyone else's.","1. Cherchez le critère, pas le rang":"1. Look for the criterion, not the rank","Le rang global est une moyenne pondérée. Un modèle peut être #1 au global et #7 en vitesse. Sur notre [[chart:ai-models]], changez de critère en haut de la page : le classement se réarrange, et c'est le but.":"The overall rank is a weighted average. A model can be #1 overall and #7 in speed. In our [[chart:ai-models]], change the criterion at the top of the page: the ranking rearranges itself, as intended.","Le rang global est une moyenne pondérée. Un modèle peut être #1 au global et #7 en vitesse. Sur notre":"The overall rank is a weighted average. A model can be #1 overall and #7 in speed. In our",", changez de critère en haut de la page : le classement se réarrange, et c'est le but.":", change the criterion at the top of the page: the ranking rearranges itself, as intended.","2. Regardez qui a mesuré":"2. Check who measured it","Un score publié par l'éditeur n'a pas la même valeur qu'un score mesuré par un tiers avec un protocole ouvert. Quand on ne peut pas faire la différence, on le dit.":"A score published by the vendor carries different weight from one measured by a third party using an open protocol. When we cannot tell the difference, we say so.","3. Regardez l'écart, pas l'ordre":"3. Look at the gap, not the order","Entre la 3ᵉ et la 4ᵉ place, il y a parfois 0,1 point. C'est du bruit. Un classement serré se lit comme un groupe, pas comme une file.":"Sometimes only 0,1 points separate 3rd and 4th place. That is noise. Read a close ranking as a group, not a queue.","4. Regardez la date":"4. Check the date","Les benchmarks vieillissent, les modèles aussi. Un score de l'an dernier dit surtout ce que l'on savait mesurer l'an dernier. Chaque page de ce site affiche sa date de mise à jour.":"Benchmarks age, and so do models. A score from last year mainly tells you what could be measured last year. Every page on this site displays its update date.","5. Regardez le prix":"5. Check the price","Deux modèles proches en qualité peuvent coûter vingt fois plus ou moins cher. C'est pour cela que la qualité globale n'inclut ni le prix ni la vitesse : les deux sont affichés à côté, pour que vous fassiez le calcul avec vos contraintes.":"Two models with similar quality can differ in cost by a factor of twenty. That is why overall quality includes neither price nor speed: both are shown alongside it so you can assess your own constraints.","Nos pondérations sont des choix éditoriaux. Elles sont publiées dans la page méthodologie de chaque classement pour pouvoir être contestées.":"Our weights are editorial choices. They are published on each ranking's methodology page so they can be challenged.","Et nous alors ?":"And what about us?","Héberger un modèle ouvert chez soi : Ollama, llama.cpp ou vLLM ?":"Hosting an open model at home: Ollama, llama.cpp or vLLM?","Trois outils, trois usages. Lequel prendre pour essayer, lequel pour un serveur, et à quel moment on regrette d'avoir choisi le mauvais.":"Three tools, three use cases. Which one to try, which one for a server, and when you regret picking the wrong one.","Les modèles du [[chart:open-source-ai]] se téléchargent. Encore faut-il les faire tourner. Trois outils reviennent tout le temps, et ils ne résolvent pas le même problème.":"Models in [[chart:open-source-ai]] can be downloaded. They still need to run. Three tools keep coming up, and they solve different problems.","Les modèles du":"Models in","se téléchargent. Encore faut-il les faire tourner. Trois outils reviennent tout le temps, et ils ne résolvent pas le même problème.":"can be downloaded. They still need to run. Three tools keep coming up, and they solve different problems.","Pour essayer : Ollama":"For trying models: Ollama","[[entity:ollama]] télécharge et lance un modèle en une commande, avec une API locale. C'est le meilleur point de départ, et pour beaucoup d'usages personnels, le seul dont on aura besoin.":"[[entity:ollama]] downloads and runs a model with one command and provides a local API. It is the best starting point and, for many personal uses, the only tool you need.","télécharge et lance un modèle en une commande, avec une API locale. C'est le meilleur point de départ, et pour beaucoup d'usages personnels, le seul dont on aura besoin.":"downloads and runs a model with one command and provides a local API. It is the best starting point and, for many personal uses, the only tool you need.","Pour tout faire tourner partout : llama.cpp":"For running models anywhere: llama.cpp","[[entity:llama-cpp]] est le moteur dont beaucoup d'autres outils dépendent. Plus bas niveau, plus de réglages, et une compatibilité matérielle très large : c'est ce qu'on prend quand on veut comprendre ou optimiser.":"[[entity:llama-cpp]] is the engine behind many other tools. It offers lower-level control, more settings and broad hardware support: use it when you want to understand or optimize how things run.","est le moteur dont beaucoup d'autres outils dépendent. Plus bas niveau, plus de réglages, et une compatibilité matérielle très large : c'est ce qu'on prend quand on veut comprendre ou optimiser.":"is the engine behind many other tools. It offers lower-level control, more settings and broad hardware support: use it when you want to understand or optimize how things run.","Pour servir plusieurs utilisateurs : vLLM":"For serving several users: vLLM","[[entity:vllm]] est pensé pour le débit : il sert beaucoup de requêtes en parallèle sur des GPU de serveur. Surdimensionné pour un usage perso, indispensable dès qu'une équipe s'y met.":"[[entity:vllm]] is designed for throughput: it handles many parallel requests on server GPUs. It is oversized for personal use but essential once a team starts using it.","est pensé pour le débit : il sert beaucoup de requêtes en parallèle sur des GPU de serveur. Surdimensionné pour un usage perso, indispensable dès qu'une équipe s'y met.":"is designed for throughput: it handles many parallel requests on server GPUs. It is oversized for personal use but essential once a team starts using it.","Vérifiez la licence du modèle (elle est sur chaque fiche), la mémoire GPU nécessaire, et si la quantification dégrade les résultats pour votre usage. Nos benchmarks portent sur les modèles non quantifiés.":"Check the model's license (shown on each profile), required GPU memory and whether quantization degrades results for your use case. Our benchmarks use unquantized models.","Avant de choisir":"Before choosing","Documentation vLLM":"vLLM documentation","Pourquoi les stars GitHub mentent à moitié, et ce qu'on en fait":"Why GitHub stars tell only half the story, and what we do about it","Une star est un signal d'attention, pas de qualité. Comment on le corrige dans nos classements : croissance, activité, échelle logarithmique.":"A star signals attention, not quality. How we account for this in our rankings: growth, activity and a logarithmic scale.","Un classement de projets par nombre de stars donne toujours les mêmes noms, ceux d'il y a cinq ans. Ce n'est pas ce qui nous intéresse : on veut savoir ce qui monte maintenant. Voici comment on s'y prend dans le [[chart:github]] et le [[chart:skills]].":"Ranking projects by total stars always produces the same names, those from five years ago. We want to know what is climbing now. Here is our approach in [[chart:github]] and [[chart:skills]].","Un classement de projets par nombre de stars donne toujours les mêmes noms, ceux d'il y a cinq ans. Ce n'est pas ce qui nous intéresse : on veut savoir ce qui monte maintenant. Voici comment on s'y prend dans le":"Ranking projects by total stars always produces the same names, those from five years ago. We want to know what is climbing now. Here is our approach in","et le":"and in","Trois corrections":"Three adjustments","**On regarde la croissance, pas le total.** Les stars gagnées sur 7 jours, et la croissance rapportée à la taille du dépôt : un petit projet qui double est visible, un géant qui grossit de 1 % ne l'est pas.":"**We look at growth, not the total.** Stars gained over 7 days and growth relative to repository size: a small project that doubles is visible; a giant growing by 1 % is not.","On regarde la croissance, pas le total.":"We look at growth, not the total.","Les stars gagnées sur 7 jours, et la croissance rapportée à la taille du dépôt : un petit projet qui double est visible, un géant qui grossit de 1 % ne l'est pas.":"Stars gained over 7 days and growth relative to repository size: a small project that doubles is visible; a giant growing by 1 % is not.","**On exige de l'activité.** Commits récents et contributeurs pèsent 25 % de la note. Un dépôt qui ne bouge plus ne peut pas rester en tête.":"**We require activity.** Recent commits and contributors account for 25 % of the score. An inactive repository cannot stay at the top.","On exige de l'activité.":"We require activity.","Commits récents et contributeurs pèsent 25 % de la note. Un dépôt qui ne bouge plus ne peut pas rester en tête.":"Recent commits and contributors account for 25 % of the score. An inactive repository cannot stay at the top.","**On écrase les extrêmes.** Les compteurs sont passés à l'échelle logarithmique : passer de 10 000 à 20 000 stars compte autant que de 100 000 à 200 000.":"**We compress the extremes.** Counts use a logarithmic scale: going from 10 000 to 20 000 stars counts as much as going from 100 000 to 200 000.","On écrase les extrêmes.":"We compress the extremes.","Les compteurs sont passés à l'échelle logarithmique : passer de 10 000 à 20 000 stars compte autant que de 100 000 à 200 000.":"Counts use a logarithmic scale: going from 10 000 to 20 000 stars counts as much as going from 100 000 to 200 000.","Ce qui reste imparfait":"What remains imperfect","Les stars peuvent être gonflées, et un pic de partages ne fait pas un projet. C'est ce qui est arrivé à [[entity:browser-use]] : en tête une partie de l'été sur un pic d'attention, puis retombé. Notre avis (OUR TAKE) sert justement à le dire, et la page de chaque projet garde la trace du pic.":"Stars can be inflated, and a spike in shares does not establish a project. This happened to [[entity:browser-use]]: leading for part of the summer after a surge in attention, then falling back. OUR TAKE explains this, and each project's page preserves the history of the peak.","Les stars peuvent être gonflées, et un pic de partages ne fait pas un projet. C'est ce qui est arrivé à":"Stars can be inflated, and a spike in shares does not establish a project. This happened to",": en tête une partie de l'été sur un pic d'attention, puis retombé. Notre avis (OUR TAKE) sert justement à le dire, et la page de chaque projet garde la trace du pic.":": leading for part of the summer after a surge in attention, then falling back. OUR TAKE explains this, and each project's page preserves the history of the peak.","Chaque classement a sa page « Comment ce classement est calculé », avec les sources, les poids et les limites. Elle se met à jour avec la méthode.":"Each ranking has a 'How this ranking is calculated' page with sources, weights and limitations. It is updated when the method changes.","Le détail des pondérations":"Weights in detail","CETTE SEMAINE":"THIS WEEK","CE MOIS":"THIS MONTH","3 MOIS":"3 MONTHS","DEPUIS LE DÉBUT":"ALL TIME","LES CLASSEMENTS IA":"THE AI CHARTS","MISE À JOUR CHAQUE JOUR":"UPDATED EVERY DAY","ÉDITION ARCHIVÉE":"ARCHIVE EDITION","Les classements quotidiens de ce qui compte dans l’IA.":"The daily charts of what matters in AI.","SEMAINE":"WEEK","À suivre":"Watchlist","LES DÉCOUVERTES":"THE DISCOVERY CHART","LE CLASSEMENT HEBDOMADAIRE":"THE WEEKLY RANKING","PARTAGER LE CLASSEMENT ↗":"SHARE CHART ↗","PARTAGER LE CLASSEMENT":"SHARE RANKING","RANG / ÉVOLUTION / PROJET":"RANK / MOVEMENT / PROJECT","DYNAMIQUE":"MOMENTUM","DONNÉES TEMPORAIREMENT INDISPONIBLES":"DATA TEMPORARILY UNAVAILABLE","BIENTÔT DISPONIBLE":"COMING NEXT","HISTORIQUE EN COURS DE CONSTITUTION":"BUILDING THE HISTORY","La collecte de ce classement se prépare.":"This ranking is being prepared.","Aucune édition publiée pour ce classement.":"No edition has been published for this ranking.","Le classement apparaîtra après la collecte des relevés et la publication d’une édition. Chaque position reposera sur des mesures enregistrées.":"The ranking will appear after observations are collected and an edition is published. Every position will be based on recorded measurements.","DANS LE PODCAST":"ON THE PODCAST","CHAQUE SEMAINE LAISSE UNE TRACE":"EVERY WEEK LEAVES A TRACE","LES ARCHIVES":"THE ARCHIVES","SANDBOX CHARTS / CHAQUE JOUR":"SANDBOX CHARTS / EVERY DAY","SUIVEZ CE QUI":"KNOW WHAT'S","BOUGE":"MOVING","Recevez les classements SANDBOX chaque jour par e-mail.":"Get SANDBOX CHARTS in your inbox every day.","RECEVOIR LES CLASSEMENTS ↗":"GET THE CHARTS ↗","RECEVOIR LES CLASSEMENTS":"GET THE CHARTS","Les projets IA qui progressent le plus vite sur GitHub.":"The fastest-moving AI projects on GitHub.","Les skills, agents et MCP qui gagnent du terrain.":"The skills, agents and MCPs gaining momentum.","Les modèles d’IA en tête actuellement.":"The models leading AI right now.","Les projets à suivre dès maintenant.":"What to watch before everyone else.","Général":"Overall","Code":"Coding","Raisonnement":"Reasoning","Mathématiques":"Maths","Contexte":"Context","Débit":"Throughput","Prix":"Price","Portée":"Reach","Poids ouverts":"Open weights","Toutes":"All","Voix":"Voice","Productivité":"Productivity","Données":"Data","Navigateur":"Browser","EN HAUSSE":"RISING","REMARQUÉ":"TRENDING","ÉTABLI":"ESTABLISHED","PLUS LONGUE SÉRIE À LA 1RE PLACE":"LONGEST #1 STREAK","PLUS FORTE HAUSSE HEBDOMADAIRE":"BIGGEST WEEKLY JUMP","PLUS DE SEMAINES DANS LE TOP 20":"MOST WEEKS IN TOP 20","PLUS FORTES HAUSSES":"BIGGEST MOVERS","Les progressions les plus nettes cette semaine.":"The biggest jumps this week.","NOUVEAUX CETTE SEMAINE":"NEW THIS WEEK","Les nouvelles entrées dans le classement.":"New entries in the charts.","EN BAISSE":"FALLING","Les projets qui reculent cette semaine.":"Losing momentum this week."," · ENTRÉE DIRECTE DANS LE TOP 10":" · DIRECT IN THE TOP 10","LA SÉLECTION DE LA RÉDACTION":"THE EDITORIAL SELECTION","À SUIVRE":"WATCHLIST","Pas encore en tête, mais déjà à suivre.":"Not big yet. We're watching them.","SUIVRE L’ATTENTION":"FOLLOW THE ATTENTION","SIGNAUX DU MARCHÉ":"MARKET SIGNALS","Variation des étoiles gagnées par les dépôts suivis,":"Change in stars earned by the tracked repositories,"," étoiles":" stars","SANDBOX CHARTS / ÉDITION MENSUELLE":"SANDBOX CHARTS / MONTHLY EDITION","Nº 1 ":"#1 ","MODÈLE":"MODEL","PROJET GITHUB":"GITHUB PROJECT","CHOIX DE SANDBOX":"SANDBOX PICK","LES RECORDS":"THE RECORD BOOK","NOUV.":"NEW","RETOUR":"RE","Dynamique des cinq derniers relevés":"Momentum over the last five observations","ÉTOILES":"STARS","SCORE SANDBOX":"SANDBOX SCORE","INDICE DU CLASSEMENT":"CHART INDEX","EN TÊTE DU CLASSEMENT":"THE ONE TO BEAT","EN TÊTE DE CETTE VUE":"LEADING THIS VIEW","CROISSANCE":"GROWTH","L’AVIS DE SANDBOX":"SANDBOX TAKE","AUCUNE ENTRÉE":"NO ENTRIES","VITESSE":"SPEED","DE QUOI S’AGIT-IL ?":"WHAT IT IS","POURQUOI ÇA MONTE ":"WHY IT'S TRENDING ","DONNÉES":"DATA","POURQUOI C’EST IMPORTANT":"WHY IT MATTERS","IDÉAL POUR":"BEST FOR","DYNAMIQUE SUR 30 JOURS":"30 DAYS MOMENTUM","Le classement, prêt à partager.":"The chart, ready to share.","Ouvrir le classement ↗":"Open the ranking ↗","Qualité générale":"Overall quality","Longueur du contexte":"Context length","Croissance des installations":"Install growth","Croissance des étoiles GitHub":"GitHub star growth","Portée des installations Skills.sh":"Skills.sh install reach","Actualité du dépôt":"Repository freshness","Dynamique entre sources":"Cross-source momentum","Vitesse des étoiles":"Star velocity","Vitesse des forks":"Fork velocity","Activité des contributeurs":"Contributor activity","Étoiles GitHub":"GitHub stars","Étoiles gagnées (7 jours)":"Stars earned (7 days)","Étoiles":"Stars","· ENTRÉE DIRECTE DANS LE TOP 10":"· DIRECT IN THE TOP 10","

Aucune présentation n’est disponible pour le moment.

":"

No presentation is available yet.

","étoiles":"stars","Nº 1":"#1","POURQUOI ÇA MONTE":"WHY IT'S TRENDING","SANDBOX CHARTS | Les classements quotidiens de l’IA":"SANDBOX CHARTS | The daily charts of AI","{count} étoiles perdues sur les relevés de cette période.":"{count} stars lost in the observations for this period.","{count} étoiles gagnées sur les relevés de cette période.":"{count} stars gained in the observations for this period.","Score SANDBOX : {score} sur 100.":"SANDBOX score: {score} out of 100.","Score de présence : {score} sur 100.":"Presence score: {score} out of 100.","{count} place gagnée depuis la période précédente.":"Up {count} place from the previous period.","{count} places gagnées depuis la période précédente.":"Up {count} places from the previous period.","{count} place perdue depuis la période précédente.":"Down {count} place from the previous period.","{count} places perdues depuis la période précédente.":"Down {count} places from the previous period.","ÉVOQUÉ DANS SANDBOX Nº":"DISCUSSED IN SANDBOX #","Mise à jour quotidienne":"Updated daily","Aperçu local. Données de démonstration.":"Local preview. Demo data.","Choisir un classement":"Choose a ranking","Cette édition":"This edition","Ce mois":"This month","3 mois":"3 months","Tout l'historique":"All time","Période":"Period","Capacité":"Capability","Options":"Options","Partager":"Share","Mode 16:9":"16:9 mode","Méthodologie et sources":"Methodology and sources","Évolution":"Change","Projet":"Project","Modèle":"Model","Skill":"Skill","Tendance":"Trend","Indice de présence aux meilleures places. Les métriques proviennent des derniers relevés disponibles.":"Index of time spent near the top. Metrics come from the latest available observations.","Positions recalculées selon la capacité choisie.":"Positions recalculated for the selected capability.","Recevoir les classements par e-mail":"Get rankings by email","Historique insuffisant":"Insufficient history","Nos projets":"Our projects","Les applications, jeux et outils de l'équipe Sandbox.":"Apps, games and tools made by the Sandbox team.","Toutes les applications":"All apps","Rechercher un projet":"Search projects","Toute l'équipe":"The whole team","Catégories de projets":"Project categories","IA et agents":"AI and agents","Culture et société":"Culture and society","À vous les commandes":"Take the controls","Développer avec l'IA":"Build with AI","Vos agents. Un même espace de travail.":"Your agents. One workspace.","Claude Code, Codex et OpenCode, du bureau au téléphone.":"Claude Code, Codex and OpenCode, from desktop to phone.","Ramenez le booster à la maison.":"Bring the booster home.","Le vol retour de Super Heavy, dans votre navigateur.":"Fly Super Heavy home in your browser.","Pour petits et grands":"For kids and grown-ups","On fait une partie ?":"Ready to play?","Faits par l'équipe":"Made by the team","Des outils qui prennent forme":"Tools we are building","Aucun projet trouvé":"No projects found","Essayez un autre nom ou retirez un filtre.":"Try another name or remove a filter.","Afficher tous les projets":"Show all projects","Aperçus":"Previews","Aperçus précédents":"Previous previews","Aperçus suivants":"Next previews","Galerie défilante":"Scrollable gallery","Capture de l'application":"App screenshot","Visuel de présentation":"Presentation artwork","Présentation du projet":"Project presentation","Aperçu agrandi":"Enlarged preview","Fermer l'aperçu":"Close preview","Voir la publication originale":"View original post","Vidéo indisponible. Voir la publication originale":"Video unavailable. View original post","À propos de cette application":"About this app","Ce que vous pouvez faire":"What you can do","Créateur":"Creator","Création":"Created by","Disponibilité":"Availability","À héberger":"Self-hosted","À installer":"Installable","En préparation":"In development","Archivé":"Archived","Code public":"Public source","Voir le code":"View source","Voir la présentation":"View presentation","Découvrir le projet":"Discover the project","Site du projet":"Project website","Dernière vérification":"Last verified","Sources de cette fiche":"Sources for this listing","Du même créateur":"More from this creator","Dépôt et documentation":"Repository and documentation","Retrouver l'équipe":"Meet the team","Obtenir mon badge":"Get my badge","Badges & Awards":"Badges & Awards","Critères et installation":"Criteria and installation","Les distinctions du projet":"Project distinctions","Classement dynamique":"Dynamic ranking","Récompense historique":"Historical award","Ce projet figure dans le Top 100 GitHub AI de Sandbox.":"This project is in the Sandbox GitHub AI Top 100.","Le dernier classement est trop ancien pour afficher un badge actif.":"The latest ranking is too old to display an active badge.","Aucune édition GitHub n'est encore publiée.":"No GitHub edition has been published yet.","Les données de classement sont temporairement indisponibles.":"Ranking data is temporarily unavailable.","Ce projet ne fait pas actuellement partie du Top 100.":"This project is not currently in the Top 100.","Dernière édition :":"Latest edition:","Mesures :":"Measurements:","Le rang suit les éditions publiées. Le badge devient neutre si le projet sort du Top 100 ou si les données ont plus de 48 heures.":"The rank follows published editions. The badge becomes neutral if the project leaves the Top 100 or the data is over 48 hours old.","Attribution retirée.":"Award withdrawn.","Premier du classement Rising à la clôture de cette semaine.":"First in the Rising ranking at the close of this week.","Consulter l'édition source":"View the source edition","L'historique des Awards est temporairement indisponible.":"Award history is temporarily unavailable.","Ce projet n'a pas encore reçu de Rising Star.":"This project has not received a Rising Star yet.","Aperçu":"Preview","Fond de l'aperçu":"Preview background","Sombre":"Dark","Clair":"Light","Chargement de l'aperçu.":"Loading preview.","L'aperçu est indisponible. Réessayez dans quelques instants.":"Preview unavailable. Try again shortly.","Format d'intégration":"Embed format","URL SVG":"SVG URL","Code pour le README GitHub":"Code for your GitHub README","Code pour votre site":"Code for your website","Adresse du badge SVG":"SVG badge address","Copier le Markdown":"Copy Markdown","Copier le HTML":"Copy HTML","Copier l'URL":"Copy URL","Copier un texte de partage":"Copy sharing text","Texte de partage copié.":"Sharing text copied.","Code Markdown copié.":"Markdown copied.","Code HTML copié.":"HTML copied.","URL du badge copiée.":"Badge URL copied.","Copie indisponible. Sélectionnez le code pour le copier.":"Copy unavailable. Select the code to copy it.","Chargement des distinctions.":"Loading distinctions.","Votre place dans les classements, prête à afficher dans votre README.":"Your ranking, ready to display in your README.","Trouver mon projet":"Find my project","Deux distinctions, deux usages":"Two distinctions, two uses","Le badge GitHub AI affiche le rang officiel du projet dans la dernière édition publiée. Il est disponible pour les rangs 1 à 100. Le classement publie actuellement 20 positions : seuls ces projets peuvent donc le récupérer.":"The GitHub AI badge displays the official project rank in the latest published edition. It is available for ranks 1 to 100. The ranking currently publishes 20 positions, so only those projects can obtain it.","Son URL reste la même. Il devient neutre si le projet quitte le Top 100, si son dépôt est indisponible ou si les mesures ont plus de 48 heures. Il se réactive automatiquement lorsque les conditions sont à nouveau réunies.":"Its URL stays the same. It becomes neutral if the project leaves the Top 100, its repository is unavailable, or measurements are over 48 hours old. It reactivates automatically when the conditions are met again.","Méthode GitHub et sources":"GitHub methodology and sources","Un prix hebdomadaire pour le premier projet GitHub du classement Rising du dimanche. L'année et la semaine ISO restent inscrites sur le badge.":"A weekly award for the first GitHub project in the Sunday Rising ranking. The year and ISO week remain on the badge.","La semaine va du lundi au dimanche, dans le fuseau Europe/Paris. L'attribution attend la clôture de la semaine et la publication de l'édition du dimanche.":"The week runs from Monday to Sunday in Europe/Paris. Assignment waits for the week to close and the Sunday edition to be published.","Rising retient les projets suivis, actifs, non archivés et non exclus, avec une accélération positive et au plus 25 000 stars par défaut. Les seuils exacts sont ceux de la configuration conservée dans l'édition source.":"Rising includes tracked, active, non-archived, non-excluded projects with positive acceleration and up to 25,000 stars by default. Exact thresholds come from the configuration preserved in the source edition.","Le rang réutilise le score Rising publié : évolution GitHub à J-7 et J-14, puis contrôle OpenSSF. Les égalités sont départagées par les stars gagnées, puis par le slug du projet.":"The rank reuses the published Rising score: GitHub changes over 7 and 14 days, followed by the OpenSSF assessment. Ties are broken by stars gained, then project slug.","Une édition absente, incomplète ou sans preuve suffisante ne donne aucun lauréat. Une seule attribution est enregistrée par semaine. Sa source et sa règle sont conservées.":"A missing, incomplete or insufficiently evidenced edition produces no winner. Only one assignment is recorded per week. Its source and rule are preserved.","Ces distinctions mesurent une position et une dynamique parmi les projets suivis. Elles ne constituent ni un audit de code, ni une certification de sécurité, ni une récompense officielle de GitHub.":"These distinctions measure position and momentum among tracked projects. They are not code audits, security certifications or official GitHub awards.","Méthode Rising et sources":"Rising methodology and sources","Les Rising Stars":"Rising Stars","Rang source :":"Source rank:","Score :":"Score:","Règle d'attribution :":"Assignment rule:","Attribué le :":"Assigned on:","Aucun Rising Star n'a encore été attribué. Le premier lauréat apparaîtra après une clôture de semaine disposant des données nécessaires.":"No Rising Star has been assigned yet. The first winner will appear after a week closes with sufficient data.","Installer votre badge":"Install your badge","Ouvrez la fiche de votre projet depuis le classement GitHub.":"Open your project page from the GitHub ranking.","Dans les distinctions, ouvrez \"Obtenir mon badge\" et copiez le Markdown.":"Under distinctions, open \"Get my badge\" and copy the Markdown.","Collez ce code dans votre README.md, puis enregistrez la modification sur GitHub. Le badge renvoie vers la fiche Sandbox. Le format HTML et l'URL SVG sont aussi disponibles.":"Paste the code into your README.md and save the change on GitHub. The badge links to the Sandbox project page. HTML and the SVG URL are also available.","Aucun compte Sandbox ni accès GitHub n'est demandé. Chaque visiteur peut copier un badge public. Seules les personnes autorisées sur le dépôt peuvent modifier son README.":"No Sandbox account or GitHub access is required. Any visitor can copy a public badge. Only people authorized on the repository can edit its README.","GitHub et les autres plateformes peuvent garder une copie de l'image en cache. Une mise à jour peut donc apparaître avec un délai. La fiche Sandbox reste la référence pour le rang et la date des mesures.":"GitHub and other platforms may cache the image, so updates can appear with a delay. The Sandbox project page remains the reference for rank and measurement dates.","Sandbox Badges & Awards":"Sandbox Badges & Awards","Affichez votre rang GitHub AI avec un badge dynamique. Découvrez les Rising Stars et les critères de leurs récompenses hebdomadaires.":"Display your GitHub AI rank with a dynamic badge. Discover Rising Stars and the criteria for their weekly awards.","Cette récompense est temporairement indisponible.":"This award is temporarily unavailable.","Toutes les récompenses":"All awards","Consulter la fiche du projet":"View project page","Source et attribution":"Source and assignment","Télécharger le SVG":"Download SVG","Guide d'installation":"Installation guide","Partager la récompense":"Share the award","Texte de partage":"Share text","Lien de la récompense":"Award link","Lien de la récompense copié.":"Award link copied.","Copier le lien de la récompense":"Copy award link","Voir la récompense":"View award","Télécharger l'image de partage":"Download share image","Réessayer l'aperçu":"Retry preview","Aucune récompense pour cette période.":"No awards for this period.","Pagination des récompenses":"Awards pagination","Éditions récentes":"Recent editions","Éditions précédentes":"Earlier editions","La récompense hebdomadaire, son classement source et son badge GitHub AI.":"The weekly award, its source ranking and its GitHub AI badge.","Consultez le retrait de cette attribution et sa source historique.":"Read about this award's withdrawal and its historical source.","Récompenses hebdomadaires Sandbox.":"Sandbox weekly awards.","Un rang actuel, des récompenses datées":"A current rank and dated awards","Les lauréats":"Award winners","Récompense":"Award","Tous les classements":"All rankings","Filtrer":"Filter","Réinitialiser les filtres":"Reset filters","Période :":"Period:","Voir les lauréats":"View winners","Les semaines suivent le calendrier ISO en Europe/Paris. Les récompenses attendent la clôture de leur période. Les sources, rangs, scores et règles sont conservés ; un retrait reste visible et ne donne pas lieu à une nouvelle attribution.":"Weeks follow the ISO calendar in Europe/Paris. Awards are assigned after their period closes. Sources, ranks, scores and rules are preserved; withdrawals remain visible and do not cause a new assignment.","Une édition absente, incomplète, retirée ou sans preuve suffisante ne donne aucun lauréat. Les prix réutilisent les classements publiés sans recalculer leurs scores. Rising exige les mesures GitHub à J-7 et J-14 et le contrôle OpenSSF.":"Missing, incomplete, withdrawn or insufficiently evidenced editions produce no winner. Awards reuse published rankings without recalculating scores. Rising requires GitHub measurements at 7 and 14 days and the OpenSSF check.","Aucune récompense n'a encore été attribuée. Les lauréats apparaîtront après une clôture disposant des données nécessaires.":"No awards have been assigned yet. Winners will appear after a period closes with the required data.","Ouvrez la fiche du projet, de la skill ou du modèle depuis son classement ou sa récompense.":"Open the project, skill or model page from its ranking or award.","Ce projet n'a pas encore reçu de récompense.":"This project has not received an award yet.","Consulter la fiche du lauréat":"View winner page","Les distinctions du lauréat":"Winner awards","Aucune récompense n'a encore été attribuée à cette entité.":"This entry has not received an award yet.","Cette fiche est temporairement indisponible.":"This page is temporarily unavailable.","Dernières apparitions":"Recent appearances","Source officielle":"Official source","Empreinte des preuves :":"Evidence hash:","Places gagnées :":"Places gained:","Rang précédent :":"Previous rank:","Premières places sur la période :":"First places during the period:","Les distinctions Sandbox.":"Sandbox awards.","Classements officiels et distinctions Sandbox.":"Official rankings and Sandbox awards.","Votre rang GitHub AI et les récompenses Sandbox : catégories, progression, Rising, Skills, modèles et palmarès annuels.":"Your GitHub AI rank and Sandbox awards: categories, movement, Rising, Skills, models and annual winners.","Premier du classement Rising du dimanche, après clôture de la semaine dans le fuseau Europe/Paris.":"First in the Sunday Rising ranking, after the week closes in Europe/Paris.","Projet le mieux classé de sa catégorie dans l'édition GitHub du dimanche. Une distinction par catégorie présente dans cette édition.":"The highest-ranked project in its category in the Sunday GitHub edition. One award per category present in that edition.","Plus forte progression positive de rang entre deux éditions GitHub du dimanche consécutives. Les nouveaux entrants sont exclus ; une égalité est départagée par le rang final, puis l'identifiant stable.":"Largest positive rank gain between consecutive Sunday GitHub editions. New entries are excluded; ties use the final rank, then the stable identifier.","Projet avec le plus de premières places Rising parmi tous les dimanches du mois civil. Une égalité est départagée par le rang du dernier dimanche, puis l'identifiant stable. Un mois incomplet ne donne aucun lauréat.":"Project with the most Rising first places across every Sunday of the calendar month. Ties use the last Sunday rank, then the stable identifier. Incomplete months produce no winner.","Première skill de l'édition officielle du dimanche, corroborée par skills.sh et GitHub, après clôture de la semaine.":"First skill in the official Sunday edition, corroborated by skills.sh and GitHub, after the week closes.","Premier modèle de l'édition officielle du dimanche, corroborée par Arena et Artificial Analysis, après clôture de la semaine.":"First model in the official Sunday edition, corroborated by Arena and Artificial Analysis, after the week closes.","Un lauréat par classement : GitHub, Rising, Skills et Models. Le plus de premières places sur tous les dimanches de l'année ISO ; égalité départagée par le dernier rang, puis l'identifiant stable. Aucun prix sans historique complet.":"One winner per ranking: GitHub, Rising, Skills and Models. Most first places across every Sunday of the ISO year; ties use the last rank, then the stable identifier. No award without complete history.","Deux mises à jour prévues par jour":"Two updates scheduled per day","Mise à jour en retard. La dernière édition publiée reste consultable.":"Update overdue. The latest published edition remains available.","Éditions classées et mesures GitHub observées pour {name}.":"Ranked editions and observed GitHub measurements for {name}.","{count} éditions à ce rang":"{count} editions at this rank","{count} éditions consécutives":"{count} editions in a row","1 édition consécutive":"1 edition in a row","Rang sur {count} éditions : {ranks}":"Rank over {count} editions: {ranks}","Plus suivi cette édition":"Most followed in this edition","Première édition du classement":"First ranking edition","Projets open source, skills et modèles IA à suivre cette semaine.":"Open source projects, skills and AI models to follow this week.","éditions dans le Top 20 · meilleur rang #":"editions in the Top 20 · best rank #","éditions en tête":"editions at the top","Editions in Top 10":"Editions in Top 10","EDITIONS IN TOP 20":"EDITIONS IN TOP 20","LES CLASSEMENTS TECH & IA":"TECH & AI CHARTS","Projets open source, skills et modèles IA à suivre chaque semaine.":"Open source projects, skills and AI models to follow each week.","PLUS D’ÉDITIONS DANS LE TOP 20":"MOST EDITIONS IN TOP 20","édition":"edition","éditions":"editions","#1 sur cette édition":"#1 in this edition","En tête depuis {count} éditions":"At the top for {count} editions","SANDBOX CHARTS | Classements open source, skills et IA":"SANDBOX CHARTS | Open source, skills and AI rankings",", édition":", edition","· édition":"· edition","· historique":"· history","(lecteur vidéo)":"(video player)","↑ ou ↓ compare la position à l’édition précédente. NEW signifie que le projet n’y figurait pas, y compris en cas de retour. Une première édition n’a pas de comparaison. Les archives conservent la configuration, les composantes, les mesures brutes et la version du score.":"↑ or ↓ compares the position with the previous edition. NEW means the project was not listed there, including when it returns. The first edition has no comparison. Archives preserve the configuration, components, raw measurements and score version.","Accélération GitHub":"GitHub acceleration","Accélération sur sept jours des jeunes projets suivis, avec contrôle des pratiques d'ingénierie.":"Seven-day acceleration among tracked early-stage projects, checked against engineering practices.","ACTUALISÉ TOUTES LES 12 H":"UPDATED EVERY 12 HOURS","Adoption des skills sur sept jours, corroborée par la dynamique du dépôt GitHub qui les distribue.":"Seven-day skill adoption, corroborated by activity in the GitHub repository that distributes each skill.","après le démarrage de la collecte":"after collection begins","Arena":"Arena","Artificial Analysis":"Artificial Analysis","Aucun animateur n’est associé à votre compte.":"No host is associated with your account.","aucun pour le moment":"none yet","Aucune baisse de rang depuis l'édition précédente.":"No ranking drops since the previous edition.","Aucune hausse de rang depuis l'édition précédente.":"No ranking gains since the previous edition.","Aucune nouvelle entrée depuis l'édition précédente.":"No new entries since the previous edition.","Aucune poussée mesurée dans les dernières 24 heures. Deux relevés espacés d'au moins une fenêtre sont nécessaires.":"No surge measured in the past 24 hours. Two observations at least one window apart are required.","Calculé depuis les positions de chaque édition du mois.":"Calculated from each edition's positions during the month.","Candidat dupliqué.":"Duplicate candidate.","candidats suivis dans cette édition. Les rangs au-delà du Top":"candidates tracked in this edition. Ranks beyond the Top","Capacités générales des variantes de modèles exactement identifiées, selon préférences humaines et évaluations standardisées.":"General capabilities of precisely identified model variants, measured through human preferences and standardized evaluations.","Ce qu'on ne sait pas faire":"What we cannot do","Ce sous-score GitHub compte pour 75 %. Les 25 % restants viennent de neuf contrôles OpenSSF sélectionnés, avec au moins six résultats exploitables.":"This GitHub subscore accounts for 75%. The remaining 25% comes from nine selected OpenSSF checks, with at least six usable results.","Ce sous-score GitHub compte pour 85 %. Les 15 % restants viennent de neuf contrôles OpenSSF sélectionnés, avec au moins six résultats exploitables.":"This GitHub subscore accounts for 85%. The remaining 15% comes from nine selected OpenSSF checks, with at least six usable results.","cette édition":"this edition","CHAQUE ÉDITION LAISSE UNE TRACE":"EVERY EDITION LEAVES A TRACE","Chaque édition v3 conserve le profil, les observations normalisées et leurs liens, les contributions, la couverture, les exclusions et une empreinte SHA-256. Les rangs de sensibilité viennent de variations de 20 % et du retrait de chaque source ou famille. Ils décrivent des scénarios, sans probabilité statistique.":"Each v3 edition preserves its profile, normalized observations and links, contributions, coverage, exclusions and a SHA-256 fingerprint. Sensitivity ranks come from 20% weight variations and removing each source or family. They describe scenarios, not statistical probabilities.","Choisir une édition de ce classement":"Choose an edition of this ranking","Classements GitHub de projets open source, skills et modèles IA. Popularité, progression, qualité et méthodes publiques.":"GitHub rankings of open-source projects, skills and AI models. Popularity, growth, quality and public methods.","Clip":"Clip","Collez la description de l’épisode avec ses sources.":"Paste the episode description and its sources.","Comment ce classement est calculé":"How this ranking is calculated","Comparaisons humaines avec contrôle du style et au moins 1000 votes. Le recrutement des votants et la sélection des variantes peuvent créer un biais.":"Human comparisons with style controls and at least 1,000 votes. Voter recruitment and variant selection may introduce bias.","Compteurs contradictoires au même instant.":"Conflicting counts at the same time.","Créneau de collecte invalide.":"Invalid collection slot.","D'où viennent les chiffres":"Where the numbers come from","Dart":"Dart","Date de collecte invalide.":"Invalid collection date.","DE QUOI S'AGIT-IL ?":"WHAT IS THIS?","dépôts suivis":"tracked repositories","Dernier rang dans les scénarios de pondération":"Lowest rank across weighting scenarios","Dernier relevé réussi :":"Last successful observation:","Désaccord entre les familles":"Disagreement between families","DÉTECTION INTRAJOURNALIÈRE":"INTRADAY DETECTION","Données absentes et couverture":"Missing data and coverage","Durée YouTube indisponible.":"YouTube duration unavailable.","Dynamique du dépôt":"Repository activity","Dynamique GitHub":"GitHub activity","ÉDITION":"EDITION","édition AAAA-Wnn ou créneau UTC":"edition YYYY-Www or UTC slot","ÉDITION ACTUELLE":"CURRENT EDITION","Edition history":"Edition history","éditions dans le Top · pic n°":"editions in the Top · peak no.","Essayez une autre catégorie ou une autre édition.":"Try another category or edition.","Évaluations standardisées":"Standardized evaluations","Évolution du nombre de contributeurs si les deux relevés existent. Cette collecte facultative n'est pas encore connectée.":"Change in contributor count when both observations are available. This optional collection is not connected yet.","Famille dupliquée.":"Duplicate family.","Famille sans source.":"Family has no source.","Fenêtre dupliquée.":"Duplicate window.","Format GitHub Trending inconnu ou liste vide.":"Unknown GitHub Trending format or empty list.","GitHub burst panel unavailable:":"GitHub burst panel unavailable:","Go":"Go","Historique :":"History:","Historique des éditions":"Edition history","Hors Top 10 cette édition":"Outside the Top 10 in this edition","Installations de skills":"Skill installs","Installations mesurées par la télémétrie du CLI. Les installations ne prouvent ni exécution, ni réussite du skill.":"Installs measured through CLI telemetry. Installs do not prove that a skill was run or worked.","Java":"Java","Jeu de mesures compté plusieurs fois.":"Measurement set counted more than once.","jours":"days","Julia":"Julia","Kotlin":"Kotlin","L'activité et la croissance mesurent l'attention des développeurs. Les stars restent manipulables et ne mesurent pas une qualité universelle.":"Activity and growth measure developer attention. Stars can be manipulated and are not a universal measure of quality.","L'avis de l'équipe, signé":"The team's signed assessment","L'AVIS DE SANDBOX":"SANDBOX'S TAKE","La collecte horaire attend sa mise en service.":"Hourly collection is awaiting activation.","La collecte n'est pas encore activée sur cet environnement. Les rangs apparaîtront après la première édition publiée.":"Collection is not enabled in this environment yet. Rankings will appear after the first edition is published.","La configuration publiée est momentanément indisponible. Réessayez après le rétablissement des données.":"The published configuration is temporarily unavailable. Try again when the data service is back.","La date et la durée doivent venir de YouTube. Configurez YOUTUBE_API_KEY sur le serveur.":"The date and duration must come from YouTube. Configure YOUTUBE_API_KEY on the server.","La découverte couvre plusieurs domaines et langues, avec des partitions de recherche. Le classement exige une licence open source reconnue par notre liste publiée. GitHub Trending sert uniquement à découvrir des candidats. Les poussées intrajournalières sont des signaux distincts, fondés sur nos propres relevés.":"Discovery covers multiple domains and languages through partitioned searches. Ranking requires an open-source license recognized by our published list. GitHub Trending is used only to discover candidates. Intraday surges are separate signals based on our own observations.","La preuve sera accessible après publication de la première édition v3.":"The evidence will be available after the first v3 edition is published.","La vidéo doit durer au moins une minute.":"The video must be at least one minute long.","Le classement apparaîtra après la collecte des relevés et la publication d'une édition. Chaque position reposera sur des mesures enregistrées.":"The ranking will appear after observations are collected and an edition is published. Every position will be based on recorded measurements.","Le classement croise aussi un rapport OpenSSF Scorecard daté de quatorze jours au plus. Un dépôt sans ce second avis peut bloquer le gel s'il pourrait entrer dans le Top 20. Ce rapport décrit des pratiques détectables, sans garantir la sécurité du code.":"The ranking also includes an OpenSSF Scorecard report no more than 14 days old. A repository without this second signal may block publication if it could enter the Top 20. The report describes detectable practices; it does not guarantee code security.","LE CLASSEMENT SANDBOX":"THE SANDBOX RANKING","Le lecteur YouTube est incomplet.":"The YouTube player is incomplete.","Le lecteur YouTube ne peut pas être chargé.":"The YouTube player could not be loaded.","Le lecteur YouTube ne répond pas.":"The YouTube player is not responding.","Le pool surveillé est limité par les quotas. Un pic ne prouve ni la qualité du code ni l'absence de manipulation. Une avance sur Trending doit se mesurer à partir de ses présences observées.":"The monitored pool is limited by quotas. A spike proves neither code quality nor the absence of manipulation. A lead over Trending must be measured from observed appearances.","Le score est la moyenne pondérée des familles, moins 20 % de leur déviation absolue pondérée. Un même jeu de mesures republié par un autre site ne reçoit pas une seconde voix.":"The score is the weighted mean of source families, minus 20% of their weighted mean absolute deviation. The same measurements republished by another site do not count as a second vote.","Lecteur YouTube de vérification de la durée":"YouTube player for verifying duration","Lecture de la durée depuis le lecteur YouTube…":"Reading duration from the YouTube player…","Les avis de l'équipe restent séparés des scores. Une correction de méthode reçoit une nouvelle version ; les éditions publiées restent immuables.":"Team assessments remain separate from scores. Method changes receive a new version; published editions stay immutable.","Les budgets sont fixes pour tous les candidats. Dans chaque famille, la puissance relative d'une source combine sa pertinence, sa rigueur, sa reproductibilité et sa résistance à la manipulation. Ce sont des choix éditoriaux à calibrer sur des éditions réelles.":"Budgets are fixed for all candidates. Within each family, a source's relative weight combines relevance, rigor, reproducibility and resistance to manipulation. These editorial choices must be calibrated against real editions.","Les nouveaux candidats GitHub et Skills constituent d'abord leurs baselines sur sept jours, Rising sur deux périodes. Avant l'admission officielle, les poussées intrajournalières peuvent déjà les signaler. Une baseline encore absente après cette attente peut bloquer le classement.":"New GitHub and Skills candidates first build a seven-day baseline; Rising candidates need two periods. Intraday surges can flag them before formal admission. A missing baseline after that wait may block publication.","Les poids des familles":"Family weights","Les poids des familles doivent totaliser 1.":"Family weights must add up to 1.","Les progressions les plus nettes depuis l'édition précédente.":"The strongest gains since the previous edition.","Les projets open source dont l'accélération mesurée mérite votre attention.":"Open-source projects whose measured acceleration deserves your attention.","Les projets open source qui progressent le plus vite sur GitHub.":"The fastest-growing open-source projects on GitHub.","Les projets qui reculent depuis l'édition précédente.":"Projects that have fallen since the previous edition.","Les projets surveillés qui gagnent rapidement des étoiles, triés par gain horaire observé. Ce signal alimente le suivi des candidats ; le classement Rising conserve ses critères et son historique.":"Monitored projects gaining stars quickly, sorted by observed hourly gain. This signal feeds candidate tracking; the Rising ranking keeps its own criteria and history.","Les relevés récents sont momentanément indisponibles.":"Recent observations are temporarily unavailable.","Les sources classantes":"Ranking sources","Les vues mensuelles et sur trois mois mesurent la présence aux meilleures places : moyenne de 100 / rang sur les éditions disponibles, avec zéro point les éditions hors classement. Elles ne moyennent pas des scores normalisés de groupes différents. Une édition datée appartient à son mois. Une semaine ISO appartient au mois de son lundi.":"Monthly and three-month views measure appearances near the top: the average of 100 divided by rank across available editions, with zero points for editions outside the ranking. They do not average normalized scores across different groups. A dated edition belongs to its calendar month. An ISO week belongs to the month containing its Monday.","Lua":"Lua","Meilleur rang dans les scénarios de pondération":"Best rank across weighting scenarios","Mesure non finie.":"Measurement is not finite.","Mesurer les projets IA qui gagnent le plus de terrain sur GitHub en sept jours.":"Measure which AI projects are gaining the most ground on GitHub over seven days.","Méthode du classement":"Ranking methodology","Minimum supérieur à la taille du classement.":"Minimum exceeds the ranking size.","Mouvements depuis l'édition précédente":"Movements since the previous edition.","Neuf contrôles de pratiques, dont six renseignés au minimum. Ce signal ne constitue pas un audit de sécurité du code.":"Nine practice checks, with at least six populated. This signal is not a code security audit.","NOUVELLES ENTRÉES":"NEW ENTRIES","Observation inconnue ou dupliquée.":"Unknown or duplicate observation.","Observation requise absente.":"Required observation is missing.","OpenSSF Scorecard":"OpenSSF Scorecard","Page Trending trop volumineuse.":"Trending page is too large.","PLUS FORTE HAUSSE ENTRE ÉDITIONS":"BIGGEST GAIN BETWEEN EDITIONS","Position d'édition en édition":"Position from edition to edition","POURQUOI C'EST IMPORTANT":"WHY IT MATTERS","Poussées d'étoiles":"Star surges","Pratiques d'ingénierie":"Engineering practices","Préférences humaines":"Human preferences","Première édition possible :":"Earliest possible edition:","Profil v3 expérimental en préparation. Les éditions antérieures conservent leur méthode.":"Experimental v3 profile in preparation. Earlier editions keep their original method.","Progression sur sept jours des dépôts open source suivis, corroborée par leurs pratiques d'ingénierie.":"Seven-day growth among tracked open-source repositories, corroborated by engineering practices.","Projets open source, skills et modèles IA à suivre dans les classements actuels.":"Open-source projects, skills and AI models to follow in the current rankings.","Projets open source, skills et modèles IA, classés et analysés deux fois par jour.":"Open-source projects, skills and AI models, ranked and analyzed twice a day.","Python":"Python","Reference methodology unavailable:":"Reference methodology unavailable:","Relevés de classement des 30 derniers jours":"Ranking observations from the past 30 days","Relevés du jour :":"Today's observations:","Repérer les projets dont la vitesse augmente avant qu'ils deviennent massifs.":"Spot projects whose growth is accelerating before they become huge.","Résumé de l'édition":"Edition summary","Ruby":"Ruby","Rust":"Rust","Seuil exceptionnel inférieur au gain minimal.":"Exceptional threshold is below the minimum gain.","Shell":"Shell","Source dupliquée.":"Duplicate source.","Source inconnue.":"Unknown source.","Source sans famille connue.":"Source has no known family.","Sources, score, critères et limites des éditions du classement GitHub.":"Sources, score, criteria and limitations for GitHub ranking editions.","Stars gagnées et récence du dépôt. Plusieurs skills peuvent partager ce signal, qui ne décrit pas leur qualité individuelle.":"Stars gained and repository recency. Several skills may share this signal; it does not describe each skill's individual quality.","Suite structurée de tâches, majoritairement textuelles et anglophones. La version du benchmark et la configuration du modèle doivent être contrôlées.":"Structured task suite, mostly text-based and in English. The benchmark version and model configuration must be checked.","SUIVRE L'ATTENTION":"TRACK ATTENTION","Swift":"Swift","Télécharger la preuve JSON de cette édition":"Download this edition's JSON evidence","Une source absente, trop ancienne ou insuffisamment renseignée ne vaut pas zéro. Toutes les sources du profil sont requises pour attribuer un rang. Les percentiles sont calculés dans le pool observé de chaque source avant de croiser les candidats. Les candidats incomplets conservent une borne haute : si elle peut atteindre le Top, le gel attend.":"A missing, stale or incomplete source does not count as zero. Every source in the profile is required to assign a rank. Percentiles are calculated within each source's observed pool before candidates are combined. Incomplete candidates retain an upper bound: if it could reach the Top, publication waits.","Unlicense":"Unlicense","UTC":"UTC","Vérifier et reproduire":"Verify and reproduce","Voir une édition":"View an edition","YouTube n’a pas fourni la date ou la durée de cette vidéo.":"YouTube did not provide this video's date or duration.","Zlib":"Zlib","Chaque rang renvoie à ses mesures, ses sources et sa méthode.":"Every rank links to its measurements, sources and methodology.","Actualisé toutes les 12 h":"Updated every 12 hours","Aucun modèle à poids ouverts dans cette édition.":"No open-weight model in this edition.","Cette mesure est indisponible pour cette édition.":"This measurement is unavailable for this edition.","Sur une période longue, le score mesure la présence aux meilleures places. Les scores de méthodes différentes ne sont pas moyennés.":"Over longer periods, the score measures presence in the top positions. Scores from different methods are not averaged.","Score pondéré de référence":"Weighted reference score","Score de cette édition":"Score for this edition","Le score combine les sources du profil figé de cette édition. Les pondérations sont des choix éditoriaux ; elles ne prouvent pas une qualité universelle.":"The score combines the sources in this edition's frozen profile. Weights are editorial choices and do not prove universal quality.","Données personnelles et mesure d'audience":"Personal data and audience measurement","Mise à jour : 14 octobre 2026":"Updated: 14 October 2026","Qui est responsable ?":"Who is responsible?","L'équipe du podcast Sandbox est responsable des traitements décrits ici. Pour une question ou l'exercice de vos droits, écrivez à":"L'équipe du podcast Sandbox is responsible for the processing described here. For questions or to exercise your rights, write to","Votre choix":"Your choice","La mesure d'audience repose sur votre consentement. Accepter et refuser sont proposés au même niveau. Le site reste accessible en cas de refus. Aucun script de mesure n'est chargé avant votre accord.":"Audience measurement relies on your consent. Accept and reject are offered at the same level. The site remains accessible if you reject. No measurement script loads before you agree.","Votre choix est enregistré dans le stockage local de ce navigateur pendant 180 jours, pour l'accord comme pour le refus. Il est ensuite redemandé. Effacer ce stockage ou changer de navigateur vous fera également refaire ce choix.":"Your choice is saved in this browser's local storage for 180 days, whether you accept or reject. You are then asked again. Clearing this storage or changing browsers also requires a new choice.","Vous pouvez retirer votre accord à tout moment avec les préférences ci-dessous ou le bouton présent en bas de chaque page. Le retrait arrête les nouveaux envois et supprime les cookies GA4 accessibles au site. Il ne supprime pas automatiquement les événements déjà reçus.":"You can withdraw consent at any time using the preferences below or the button at the bottom of each page. Withdrawal stops new transmissions and deletes GA4 cookies accessible to the site. It does not automatically delete events already received.","Ce qui est mesuré après accord":"What is measured after consent","Google Analytics 4 mesure les pages consultées et les actions utiles, comme l'ouverture du lecteur ou d'une ressource. Les événements peuvent inclure le chemin de la page, son titre, le site d'origine, la date, le type d'appareil, le navigateur et une localisation approximative. Les paramètres de recherche et fragments d'URL sont retirés de nos envois ; aucun nom, e-mail ou texte saisi dans la recherche n'est envoyé par nos événements.":"Google Analytics 4 measures visited pages and useful actions, such as opening the player or a resource. Events may include the page path, title, referring site, date, device type, browser and approximate location. Search parameters and URL fragments are removed from our transmissions; our events send no name, email address or text entered in search.","Vercel Web Analytics produit également des statistiques de fréquentation. Vercel indique ne pas utiliser de cookies pour ce service et supprimer sous 24 heures l'identifiant calculé pour distinguer les visites. Les statistiques agrégées restent consultables selon l'offre Vercel, de 1 à 24 mois. Nos envois excluent les paramètres d'URL et la zone d'administration.":"Vercel Web Analytics also produces traffic statistics. Vercel states that it uses no cookies for this service and discards the identifier calculated to distinguish visits within 24 hours. Aggregated statistics remain available for 1 to 24 months depending on the Vercel plan. Our transmissions exclude URL parameters and the administration area.","Nous désactivons les signaux Google et la personnalisation publicitaire dans le tag du site. Nous n'envoyons pas d'identifiant utilisateur ni de données de compte dans nos événements.":"We disable Google signals and ad personalization in the site's tag. Our events send no user ID or account data.","Prestataires et services externes":"Providers and external services","Google et Vercel reçoivent les données nécessaires à leurs services de mesure. Leur traitement peut avoir lieu hors de l'Union européenne. Leurs politiques décrivent les lieux de traitement, les destinataires et les garanties de transfert. Vous pouvez refuser ces services ou retirer votre accord.":"Google and Vercel receive the data needed for their measurement services. Processing may take place outside the European Union. Their policies describe processing locations, recipients and transfer safeguards. You can reject these services or withdraw consent.","Politique de confidentialité de Google":"Google privacy policy","Conservation des données dans Google Analytics":"Google Analytics data retention","Données traitées par Vercel Web Analytics":"Data processed by Vercel Web Analytics","Politique de confidentialité de Vercel":"Vercel privacy policy","Le lecteur YouTube est chargé lorsque vous demandez la lecture d'un épisode. YouTube peut alors recevoir votre adresse IP et des informations de lecture, selon ses propres règles. Le choix de mesure du site ne remplace pas les paramètres de votre compte YouTube. Les liens vers les autres plateformes vous conduisent sur leurs services.":"The YouTube player loads when you request episode playback. YouTube may then receive your IP address and playback information under its own rules. The site's measurement choice does not replace your YouTube account settings. Links to other platforms take you to their services.","L'hébergement doit traiter les requêtes réseau nécessaires à l'affichage et à la sécurité du site. Les préférences d'affichage et les favoris éventuels sont conservés dans votre navigateur pour rendre les fonctions demandées ; ils ne déclenchent pas la mesure d'audience.":"Hosting processes the network requests needed to display and secure the site. Display preferences and any favorites are stored in your browser to provide the requested features; they do not activate audience measurement.","Vos droits":"Your rights","Vous pouvez demander l'accès, la rectification, l'effacement ou la limitation des données qui vous concernent, et leur portabilité lorsque ce droit s'applique. Vous pouvez retirer votre consentement sans remettre en cause les traitements réalisés avant son retrait. Contactez l'équipe à l'adresse indiquée plus haut en précisant votre demande, sans transmettre de document d'identité spontanément.":"You can request access, correction, erasure or restriction of your data, and portability where this right applies. You can withdraw consent without affecting processing carried out before withdrawal. Contact the team at the address above and describe your request, without sending identity documents unsolicited.","Vous pouvez aussi adresser une réclamation à la CNIL.":"You can also lodge a complaint with the CNIL.","Contacter la CNIL":"Contact the CNIL","Voir cette édition":"View this edition","Ce que les chiffres permettent de dire":"What the figures can tell us","Une popularité cumulée ne mesure pas une progression récente. Une mesure de benchmark dépend de ses tâches, de sa date et des modèles couverts. Les scores sont relatifs au pool observé.":"Cumulative popularity does not measure recent growth. A benchmark depends on its tasks, date and model coverage. Scores are relative to the observed pool.","Les avis de l'équipe sont signés et séparés du calcul. Une recommandation dans le podcast ne modifie pas un rang.":"The team's opinions are signed and separate from the calculation. A podcast recommendation does not change a rank.","Cette archive conserve sa version historique. La méthode v3 et sa calibration sont suivies séparément ; elles ne sont pas appliquées rétroactivement.":"This archive keeps its historical version. The v3 method and its calibration are tracked separately and are not applied retroactively.","Mesures et sources conservées":"Preserved measurements and sources","Les mesures ci-dessous sont celles de l'édition. Une métrique affichée en contexte ne contribue pas nécessairement au score. Les poids effectifs ne sont indiqués que lorsqu'ils ont été conservés.":"The measurements below belong to this edition. A metric shown for context does not necessarily contribute to the score. Effective weights are shown only where they were preserved.","Nom et sources":"Name and sources","Mesures conservées":"Preserved measurements","Poids effectifs conservés":"Preserved effective weights","Non renseignés dans cette archive":"Not recorded in this archive","Configuration historique":"Historical configuration","Les paramètres ci-dessous ont été figés lors du calcul. Ils peuvent contenir des dimensions prévues mais absentes de certaines lignes. Ils ne remplacent pas les mesures et les poids effectifs conservés.":"The parameters below were frozen at calculation time. They may contain planned dimensions absent from some rows. They do not replace the preserved measurements and effective weights.","Lire la configuration JSON":"Read the JSON configuration","Ouvrir les données JSON de cette édition":"Open this edition's JSON data","Relevé, calcul et publication":"Observation, calculation and publication","Le relevé date une observation chez une source. Le calcul fige le score avec les observations disponibles. La publication rend cette édition accessible. Ces trois dates peuvent être différentes ; une nouvelle collecte ne republie pas automatiquement un classement.":"An observation dates a measurement at a source. Calculation freezes the score using available observations. Publication makes the edition accessible. These dates can differ; a new collection does not automatically republish a ranking.","Les créneaux de calcul prévus à 03:17 et 15:17 UTC ne garantissent pas une nouvelle édition : la couverture, les baselines et la validation peuvent la retarder. La date affichée reste celle de l'édition effectivement publiée.":"Scheduled calculation slots at 03:17 and 15:17 UTC do not guarantee a new edition: coverage, baselines and validation can delay it. The displayed date belongs to the edition actually published.","Profil expérimental : pondérations en cours de calibration.":"Experimental profile: weights are being calibrated.","Codex, Claude Code, OpenCode : le modèle ne suffit pas":"Codex, Claude Code, OpenCode: the model is not enough","Que change le harnais autour du modèle ? Lou, Nicolas et Loïc présentent Sillage à 25:50, puis confrontent les approches de Codex, Claude Code et OpenCode à 37:17, avec leurs essais et leurs limites.":"What does the harness around a model change? Lou, Nicolas and Loïc introduce Sillage at 25:50, then compare the approaches of Codex, Claude Code and OpenCode at 37:17, with their experiments and limitations.","Éditions datées et sourcées":"Dated, sourced editions","Dernier relevé inclus":"Latest included observation","Calcul figé":"Frozen calculation","Publication":"Publication","Non publiée":"Unpublished","Version du calcul":"Calculation version","Version du calcul :":"Calculation version:","Données personnelles":"Personal data","Préférences de mesure":"Measurement preferences","Les projets de cet épisode":"Projects in this episode","Découvrir":"Discover","Un projet de":"A project by","Dans le podcast":"In the podcast","Méthode de l'édition":"Edition methodology","Popularité GitHub":"GitHub popularity","Popularité Skills.sh":"Skills.sh popularity","Mesures de modèles disponibles":"Available model measurements","Cette édition classe les stars cumulées. L'historique disponible ne permettait pas encore de mesurer leur progression sur sept jours.":"This edition ranks cumulative stars. The available history was not yet sufficient to measure seven-day growth.","Cette édition classe les installations cumulées de Skills.sh. Les étoiles GitHub sont un contexte, pas une seconde mesure du score.":"This edition ranks cumulative Skills.sh installations. GitHub stars provide context and do not count as a second scoring measure.","Les sources disponibles varient selon les modèles de cette édition. Un score ne garantit pas deux évaluations indépendantes ni une mesure du rapport qualité-prix.":"Available sources vary between models in this edition. A score does not guarantee two independent evaluations or a measure of value for money.","Les rangs décrivent le pool observé. Les mesures de popularité et les avis de notre équipe restent distincts.":"Ranks describe the observed pool. Popularity measurements and our team's opinions remain separate.","Les mesures de modèles se lisent avec leurs sources, leur couverture et leurs limites.":"Model measurements should be read with their sources, coverage and limitations.","GA4 utilise les cookies _ga et _ga_* pour distinguer les visites. Leur durée est limitée à 180 jours sans prolongation automatique. La durée retenue par l'équipe pour les données d'événements GA4 est de 14 mois ; le réglage effectif du flux doit être vérifié avant son activation. Cette durée ne s'applique pas de la même façon aux rapports agrégés, selon les règles de Google.":"GA4 uses _ga and _ga_* cookies to distinguish visits. They last up to 180 days without automatic renewal. Event data retention in our GA4 property is intended to be 14 months; the effective setting must be checked before activation. Under Google's rules, this period does not apply in the same way to aggregated reports.","Le podcast":"The podcast","Un épisode par semaine sur la tech et l'IA.":"A weekly episode about tech and AI.","S'abonner sur YouTube":"Subscribe on YouTube","Dernier replay":"Latest episode","Voir le replay":"Watch the replay","Épisodes précédents":"Previous episodes","Aucun replay pour le moment.":"No replays yet.","Le premier épisode sera disponible ici.":"The first episode will be available here.","Le podcast Sandbox : replays et historique":"The Sandbox podcast: replays and archive","Un épisode par semaine sur la tech et l'IA. Retrouvez le dernier replay Sandbox et les épisodes précédents.":"A weekly episode about tech and AI. Watch the latest Sandbox replay and previous episodes.","Voir les replays":"View replays","Essayez un autre mot.":"Try another word.","résultats":"results","à explorer.":"ready to explore.","L'IA et la tech,":"AI and tech,","Le terrain de jeu des curieux":"A place for curious minds","Des outils à découvrir, des projets à essayer et une histoire à comprendre.":"Discover tools, try projects and explore the history behind AI.","Archives des classements":"Ranking archives","Chronologie en français":"Timeline in French","Classements disponibles":"Available rankings","Comment en est-on arrivé là ?":"How did we get here?","Comparez les modèles selon la capacité qui vous intéresse.":"Compare models by the capability you need.","Comprendre les données de cette édition":"Understand this edition's data","Créés par l'équipe":"Built by the team","Dans l'émission":"In the show","Dans les classements":"In the rankings","Découvrir le podcast":"Discover the podcast","Découvrir nos projets":"Explore our projects","Dernier épisode":"Latest episode","Des instructions et outils pour vos agents IA.":"Instructions and tools for your AI agents.","Des outils à découvrir. Des données pour les comparer.":"Discover tools and use the data to compare them.","Et derrière SANDBOX ?":"Who's behind SANDBOX?","Explorer":"Explore","Explorer la chronologie":"Explore the timeline","Explorer les classements":"Explore the rankings","Explorez les projets open source du classement GitHub.":"Explore open-source projects in the GitHub ranking.","Guides":"Guides","Histoire de l'IA":"AI history","L'actualité IA et tech.":"AI and tech news.","Ce qu'on essaie en ce moment.":"What we're trying right now.","La Bêta":"La Bêta","Le Build":"Le Build","Le Clash":"Le Clash","Le Radar":"Le Radar","Le podcast de Lou, Loïc et Nico":"The podcast with Lou, Loïc and Nico","Les classements attendent leur prochaine publication.":"The next rankings are awaiting publication.","Les classements sont temporairement indisponibles.":"Rankings are temporarily unavailable.","Les projets repérés par le classement Rising.":"Projects highlighted by the Rising ranking.","Lou, Loïc et Nico. Trois passionnés qui se retrouvent chaque semaine pour parler d'IA et de tech.":"Lou, Loïc and Nico meet every week to talk about AI and tech.","Lou, Loïc et Nico. Un rendez-vous hebdomadaire pour parler d'IA et de tech.":"Join Lou, Loïc and Nico for a weekly conversation about AI and tech.","Mise à jour en retard. Dernière édition disponible.":"Update overdue. Showing the latest available edition.","Modèles IA":"AI models","Nos débats et nos désaccords.":"Our debates and disagreements.","Nos projets et retours d'expérience.":"Our projects and what we've learned.","Regarder sur YouTube":"Watch on YouTube","Remontez l'histoire de l'IA, une annonce à la fois. Chaque date renvoie à sa source.":"Explore AI history, one announcement at a time. Every date links to its source.","Rencontrer l'équipe":"Meet the team","Retrouvez-nous sur YouTube":"Find us on YouTube","Time Machine (en français)":"Time Machine (in French)","Un site pour explorer l'IA. Une émission pour en parler.":"A site to explore AI. A show to talk about it.","Voir l'épisode complet":"Watch the full episode","Voir le classement GitHub":"View the GitHub ranking","Deep Blue bat Kasparov":"Deep Blue beats Kasparov","L'architecture Transformer":"The Transformer architecture","L'ouverture de ChatGPT":"ChatGPT opens to the public","Sandbox":"Sandbox","SANDBOX":"SANDBOX","SANDBOX /":"SANDBOX /","Sandbox Charts":"Sandbox Charts","Time Machine /":"Time Machine /","Débutant":"Beginner","Intermédiaire":"Intermediate","Avancé":"Advanced","Le podcast IA & tech":"The AI & tech podcast","Comprendre l'IA.":"Understand AI.","En la pratiquant.":"By putting it to work.","Chaque semaine, Lou, Loïc et Nico partagent leurs essais d'outils IA et les projets qu'ils construisent avec.":"Every week, Lou, Loïc and Nico share the AI tools they try and the projects they build with them.","Le podcast IA & tech de Lou, Loïc et Nico.":"The AI & tech podcast with Lou, Loïc and Nico.","Regarder le dernier épisode":"Watch the latest episode","Découvrir l'émission":"Discover the show","Visuel de plateau provisoire":"Temporary set visual","L'IA, en pratique.":"AI, in practice.","On teste des outils. On construit des projets. On en parle ensemble.":"We try tools. We build projects. We talk it through together.","Plateau provisoire · photomontage":"Temporary set · photomontage","Décor provisoire":"Temporary set","Gardez une longueur d'avance sur la tech.":"Stay one step ahead in tech.","Chaque semaine, retrouvez Lou, Loïc et Nico pour décrypter l'actualité de l'IA et de la tech, découvrir les outils qui valent le détour et confronter nos expériences.":"Every week, join Lou, Loïc and Nico to make sense of AI and tech news, discover tools worth your time and compare our experiences.","De gauche à droite : Lou, Loïc et Nico.":"From left to right: Lou, Loïc and Nico.","La suite":"Explore","Réduire les mouvements":"Reduce motion","Activer les mouvements":"Enable motion","Derrière les micros":"Behind the microphones","Les voix de Sandbox.":"The voices of Sandbox.","Jeux, apprentissage et usages":"Games, learning and practical uses","IA locale et création musicale":"Local AI and music creation","Agents de code et expériences web":"Coding agents and web experiences","Du podcast au concret":"From podcast to projects","Ce qu'on construit.":"What we build.","Des jeux, de la musique, des outils. Voici quelques-uns de nos projets.":"Games, music, tools. Here are some of our projects.","Voir tous nos projets":"See all our projects","Pour aller plus loin":"Explore further","Explorer l'IA.":"Explore AI.","Remonter l'histoire de l'IA, de ses premières avancées aux modèles actuels.":"Trace the history of AI, from early advances to today’s models.","Explorer les projets, les modèles et les skills, avec leurs sources et leurs méthodes.":"Explore projects, models and skills, with their sources and methods.","Lou construit des applications pour jouer et apprendre. Avec Not My Fault, il met la communication d'une équipe à l'épreuve dans un jeu coopératif. Héros de la classe transforme les révisions du primaire en aventures, tandis qu'Okodukai aborde la gestion de l'argent par le jeu. Ses projets explorent différentes façons de rendre un sujet concret et accessible.":"Lou builds applications for play and learning. With Not My Fault, he puts team communication to the test in a cooperative game. Héros de la classe turns primary school revision into adventures, while Okodukai introduces money management through play. His projects explore different ways to make a subject practical and accessible.","Loïc développe des outils pour faire fonctionner l'IA sur sa propre machine. Akasha OS rassemble conversations, modèles, mémoire et agents dans un environnement local. Avec Song Maker, il explore la création musicale : génération de chansons, édition des prises et mixage. Ses projets vont des outils techniques aux applications de création.":"Loïc develops tools for running AI on your own machine. Akasha OS brings conversations, models, memory and agents together in a local environment. With Song Maker, he explores music creation: generating songs, editing takes and mixing. His projects range from technical tools to creative applications.","Nico crée des outils pour travailler avec des agents et des expériences à explorer dans le navigateur. Sillage réunit Claude Code, Codex et OpenCode dans un même espace de travail. BallCam.tv permet de revoir des matchs Rocket League en 3D, et Boostback de piloter le retour de Super Heavy. Ses réalisations relient développement, jeu et simulation.":"Nico creates tools for working with agents and experiences to explore in the browser. Sillage brings Claude Code, Codex and OpenCode into a single workspace. BallCam.tv lets you revisit Rocket League matches in 3D, while Boostback lets you pilot the return of Super Heavy. His work connects development, games and simulation.","Jeux vidéo, éducation et apprentissage":"Video games, education and learning","Passionné par les jeux vidéo et les nouvelles technologies, Lou explore comment appliquer les mécaniques du jeu vidéo à l'éducation et à la pédagogie. Son ambition : transformer les programmes scolaires et les apprentissages du quotidien en expériences ludiques et interactives, pour apprendre autrement, en mêlant l'utile à l'agréable.":"Passionate about video games and new technologies, Lou explores how to apply game mechanics to education and teaching. His ambition: turn school curricula and everyday learning into fun, interactive experiences, combining enjoyment with practical value to learn in new ways.","À travers des projets comme Héros de la classe, Okodukai ou Not My Fault, il expérimente de nouvelles façons d'apprendre, de collaborer et de développer ses compétences en jouant.":"Through projects such as Héros de la classe, Okodukai and Not My Fault, he experiments with new ways to learn, collaborate and develop skills through play.","Game designer des trucs chiants":"Game designer for the boring bits","Savant fou en localhost":"Mad scientist on localhost","Directeur de la flemme automatisée":"Chief automated laziness officer","Son terrain de jeu : le local et l'open source. Il veut comprendre ce qui tourne sous le capot, assembler ses propres outils et garder la main sur ses propres machines, au quotidien.":"His playground: local tools and open source. He wants to understand what runs under the hood, build his own tools and stay in control of his own machines, every day.","Son autre obsession : optimiser et automatiser le quotidien. Il cherche ce qu'il peut confier aux IA et aux robots, pour consacrer davantage de temps aux idées suivantes.":"His other obsession: optimizing and automating everyday life. He looks for tasks he can hand over to AI and robots, leaving more time for whatever idea comes next.","Nico crée des outils pour travailler avec des agents et des expériences à explorer dans le navigateur. Sillage réunit Claude Code, Codex et OpenCode dans un même espace de travail. BallCam.tv permet de revoir des matchs Rocket League en 3D, et Boostback de piloter le retour de Super Heavy. Ses réalisations relient développement, jeu et simulation.\n\nSon autre obsession : optimiser et automatiser le quotidien. Il cherche ce qu'il peut confier aux IA et aux robots, pour consacrer davantage de temps aux idées suivantes.":"Nico creates tools for working with agents and experiences to explore in the browser. Sillage brings Claude Code, Codex and OpenCode into a single workspace. BallCam.tv lets you revisit Rocket League matches in 3D, while Boostback lets you pilot the return of Super Heavy. His work connects development, games and simulation.\n\nHis other obsession: optimizing and automating everyday life. He looks for tasks he can hand over to AI and robots, leaving more time for whatever idea comes next.","Nico crée des outils pour travailler avec des agents et des expériences à explorer dans le navigateur. Sillage réunit Claude Code, Codex et OpenCode dans un même espace de travail. BallCam.tv permet de revoir des matchs Rocket League en 3D, et Boostback de piloter le retour de Super Heavy. Ses réalisations relient développement, jeu et simulation. Son autre obsession : optimiser et automatiser le quotidien. Il cherche ce qu'il peut confier aux IA et aux robots, pour consacrer davantage de temps aux idées suivantes.":"Nico creates tools for working with agents and experiences to explore in the browser. Sillage brings Claude Code, Codex and OpenCode into a single workspace. BallCam.tv lets you revisit Rocket League matches in 3D, while Boostback lets you pilot the return of Super Heavy. His work connects development, games and simulation. His other obsession: optimizing and automating everyday life. He looks for tasks he can hand over to AI and robots, leaving more time for whatever idea comes next.","Loïc développe des outils pour faire fonctionner l'IA sur sa propre machine. Akasha OS rassemble conversations, modèles, mémoire et agents dans un environnement local. Avec Song Maker, il explore la création musicale : génération de chansons, édition des prises et mixage. Ses projets vont des outils techniques aux applications de création.\n\nSon terrain de jeu : le local et l'open source. Il veut comprendre ce qui tourne sous le capot, assembler ses propres outils et garder la main sur ses propres machines, au quotidien.":"Loïc develops tools for running AI on your own machine. Akasha OS brings conversations, models, memory and agents together in a local environment. With Song Maker, he explores music creation: generating songs, editing takes and mixing. His projects range from technical tools to creative applications.\n\nHis playground: local tools and open source. He wants to understand what runs under the hood, build his own tools and stay in control of his own machines, every day.","Loïc développe des outils pour faire fonctionner l'IA sur sa propre machine. Akasha OS rassemble conversations, modèles, mémoire et agents dans un environnement local. Avec Song Maker, il explore la création musicale : génération de chansons, édition des prises et mixage. Ses projets vont des outils techniques aux applications de création. Son terrain de jeu : le local et l'open source. Il veut comprendre ce qui tourne sous le capot, assembler ses propres outils et garder la main sur ses propres machines, au quotidien.":"Loïc develops tools for running AI on your own machine. Akasha OS brings conversations, models, memory and agents together in a local environment. With Song Maker, he explores music creation: generating songs, editing takes and mixing. His projects range from technical tools to creative applications. His playground: local tools and open source. He wants to understand what runs under the hood, build his own tools and stay in control of his own machines, every day.",", histoire mondiale de l'IA":", world history of AI",". Aucun score de performance.":". No performance score.",". Dernier événement :":". Latest event:",". GitHub AI.":". GitHub AI.","\">":"","","ANNUAL AWARDS":"ANNUAL AWARDS","Anthropic annonce Claude, son assistant conversationnel issu de recherches sur l'alignement.":"Anthropic announces Claude, its conversational assistant developed from alignment research.","Anthropic annonce la famille Claude 3 et rend Opus et Sonnet accessibles dans Claude et via son API.":"Anthropic announces the Claude 3 family and makes Opus and Sonnet available in Claude and through its API.","Anthropic annonce sa Cyber Mission":"Anthropic announces its Cyber Mission","Anthropic lance Claude 3.5 Sonnet":"Anthropic launches Claude 3.5 Sonnet","Anthropic lance Claude Fable 5 et Mythos 5":"Anthropic launches Claude Fable 5 and Mythos 5","Anthropic lance Claude Opus 4 et Sonnet 4, orientés vers le code, le raisonnement et les tâches longues.":"Anthropic launches Claude Opus 4 and Sonnet 4, focused on coding, reasoning and long tasks.","Anthropic lance Claude Opus 5.5":"Anthropic launches Claude Opus 5.5","Anthropic lance Fable 5 et Mythos 5, deux déclinaisons d'un même modèle avec des protections différentes.":"Anthropic launches Fable 5 and Mythos 5, two versions of the same model with different safeguards.","Anthropic lance Haiku 5.5 pour les tâches rapides et répétitives, notamment les résumés, la classification et les sous-agents de code.":"Anthropic launches Haiku 5.5 for fast, repetitive tasks, including summaries, classification and coding subagents.","Anthropic lance Sonnet 5.5 avec des améliorations annoncées en rapidité et en coût sur les tâches professionnelles et le code.":"Anthropic launches Sonnet 5.5 with announced speed and cost improvements for professional tasks and coding.","Anthropic présente Claude":"Anthropic introduces Claude","Anthropic présente Opus 5.5 pour les travaux complexes, avec des capacités rapprochées de Fable 5.1 sur ses évaluations.":"Anthropic introduces Opus 5.5 for complex work, with capabilities close to Fable 5.1 in its evaluations.","Anthropic présente une initiative consacrée à la défense informatique et aux usages de ses modèles par les équipes de sécurité.":"Anthropic presents an initiative focused on cyberdefense and security teams' use of its models.","Anthropic publie Claude 3.7 Sonnet, qui associe réponses rapides et raisonnement prolongé.":"Anthropic releases Claude 3.7 Sonnet, combining quick answers with extended reasoning.","Anthropic publie le Model Context Protocol pour relier les assistants IA aux outils et aux sources de données.":"Anthropic releases the Model Context Protocol to connect AI assistants to tools and data sources.","Apple annonce l'intégration de Siri à l'iPhone 4S pour effectuer des actions et répondre à des demandes vocales.":"Apple announces Siri integration into the iPhone 4S to carry out actions and respond to voice requests.","Apprendre à gérer son argent en jouant.":"Learn to manage money through play.","Article original et métadonnées CVPR 2014":"Original paper and CVPR 2014 metadata","Article original sur BERT":"Original BERT paper","Article original sur Imagen":"Original Imagen paper","Article original sur T5":"Original T5 paper","Article original sur word2vec":"Original word2vec paper","Article original, CVPR 2009":"Original paper, CVPR 2009","Article original, Neural Computation":"Original paper, Neural Computation","Article original, NIPS 2012":"Original paper, NIPS 2012","Article original, Proceedings of the IEEE":"Original paper, Proceedings of the IEEE","Artificial Analysis publie une première évaluation de modèles généralistes qui pilotent un bras robotique sur huit tâches.":"Artificial Analysis publishes an initial evaluation of general-purpose models controlling a robotic arm across eight tasks.","Attention Is All You Need":"Attention Is All You Need","Attention Is All You Need présente une architecture de traitement des séquences fondée sur des mécanismes d'attention.":"Attention Is All You Need presents an attention-based architecture for sequence processing.","au mois":"to month","Aucun événement sélectionné.":"No event selected.","Aucun résultat pour cette sélection.":"No results for this selection.","Aucun résultat.":"No results.","Aucune cassette trouvée.":"No tapes found.","Aucune édition publique valide.":"No valid public edition.","Audience measurement":"Audience measurement","Auto-Encoding Variational Bayes":"Auto-Encoding Variational Bayes","AWARD WITHDRAWN":"AWARD WITHDRAWN","Badges & Awards":"Badges & Awards","BallCam transforme les fichiers de replay Rocket League en matchs à explorer en 3D dans le navigateur. Vous pouvez revoir une action, changer de point de vue et partager un replay avec votre équipe ou votre coach.":"BallCam turns Rocket League replay files into matches to explore in 3D in the browser. You can revisit a play, change viewpoints and share a replay with your team or coach.","BERT apprend le contexte dans les deux sens":"BERT learns context in both directions","BERT préentraîne un Transformer bidirectionnel, puis l'adapte à différentes tâches de compréhension du langage.":"BERT pretrains a bidirectional Transformer, then adapts it to various language understanding tasks.","Best AI Model":"Best AI Model","BEST AI MODEL":"BEST AI MODEL","BIENVENUE AU CLUB":"WELCOME TO THE CLUB","Biggest Mover":"Biggest Mover","BIGGEST MOVER":"BIGGEST MOVER","Bilan de la rencontre, Google DeepMind":"Match recap, Google DeepMind","Bilan officiel de la première année de PyTorch":"Official recap of PyTorch's first year","Black Forest Labs publie FLUX.1":"Black Forest Labs releases FLUX.1","Black Forest Labs publie FLUX.1, une famille de modèles texte-vers-image en trois variantes.":"Black Forest Labs releases FLUX.1, a family of text-to-image models in three variants.","BLOOM ouvre un grand modèle multilingue à la recherche":"BLOOM opens a large multilingual model to research","Boostback est un jeu de pilotage spatial en 3D. Vous prenez les commandes du booster Super Heavy pour le ramener vers Starbase et tenter de l'attraper avec les bras de Mechazilla.":"Boostback is a 3D spaceflight game. Take control of the Super Heavy booster to bring it back to Starbase and try to catch it with Mechazilla's arms.","Bot Chronicles est un blog écrit par deux agents IA, Marlbot et Pelouse. Ils utilisent Bot Hub pour choisir leurs sujets et rédiger des articles sur leurs expériences d'agents et leur infrastructure.":"Bot Chronicles is a blog written by two AI agents, Marlbot and Pelouse. They use Bot Hub to choose topics and write articles about their experiences as agents and their infrastructure.","Bot Hub permet à plusieurs bots OpenClaw de discuter entre eux. Chaque bot rejoint des salons et conserve une session de conversation dédiée à chaque salon.":"Bot Hub lets multiple OpenClaw bots talk to one another. Each bot joins rooms and maintains a separate conversation session for each room.","Calcul, données et premières victoires":"Compute, data and early victories","Calendrier officiel des cinq matchs":"Official schedule of the five matches","Canonical différent de l'URL annoncée":"Canonical differs from the stated URL","Canonical unique attendu":"A unique canonical URL is required","Cassette":"Tape","cassettes":"tapes","Catalogue classique":"Classic catalogue","Catégorie inconnue.":"Unknown category.","Ce firmware transforme une carte Freenove ESP32-S3 Display FNK0104B en compagnon vocal du daemon Akasha. Le microphone et le haut-parleur servent à dialoguer sur le réseau local.":"This firmware turns a Freenove ESP32-S3 Display FNK0104B board into a voice companion for the Akasha daemon. The microphone and speaker allow conversations over the local network.","Ce service maintient les enregistrements DNS Cloudflare à partir des labels des conteneurs Docker et des services Swarm. Il observe leurs événements pour créer ou mettre à jour les entrées correspondantes.":"This service maintains Cloudflare DNS records from Docker container and Swarm service labels. It watches their events to create or update matching records.","Cette distinction ne prend pas de catégorie.":"This award does not accept a category.","Chaîne de collecte invalide.":"Invalid collection chain.","Chaque fiche distingue annonce, publication et sortie. Une annonce ne garantit pas un accès public. Les dates conservent leur précision d'origine.":"Each entry distinguishes announcement, publication and release. An announcement does not guarantee public access. Dates retain their original precision.","Chaque monde correspond à une période de l'Histoire. Les missions mobilisent lecture, calcul, logique et connaissances historiques. La page publique présente le projet, construit par deux pères de famille, et permet de rejoindre la liste des familles intéressées par la bêta.":"Each world corresponds to a historical period. Missions involve reading, arithmetic, logic and historical knowledge. The public page introduces the project, built by two fathers, and lets families join the beta interest list.","Chaque projet possède un tableau de tâches, ses fichiers, ses branches Git et l'historique des échanges. Les sessions peuvent retrouver les décisions précédentes, partager des instructions et laisser du contexte aux autres agents. Les tâches planifiées démarrent de nouvelles sessions selon la cadence choisie.":"Each project has a task board, files, Git branches and conversation history. Sessions can retrieve earlier decisions, share instructions and leave context for other agents. Scheduled tasks start new sessions at the chosen frequency.","Chargement de la chronologie mondiale de l'IA.":"Loading the world history of AI timeline.","ChatGPT est lancé en préversion de recherche gratuite avec une interface de dialogue permettant les questions successives.":"ChatGPT launches as a free research preview with a conversational interface supporting follow-up questions.","ChatGPT intègre les capacités de navigation d'Operator":"ChatGPT integrates Operator's browsing capabilities","ChatGPT ouvre l'IA conversationnelle au public":"ChatGPT opens conversational AI to the public","ChatGPT Space réunit équipes, documents et agents":"ChatGPT Space brings teams, documents and agents together","chemin local attendu, ex. /team/nom.jpg":"local path expected, e.g. /team/nom.jpg","Chinchilla rééquilibre paramètres et données":"Chinchilla rebalances parameters and data","Choisir son niveau":"Choose your level","Choisir son vol":"Choose your flight","Choisir un lieu sur la carte":"Choose a location on the map","Choisis une boîte. Le projet est à l'intérieur.":"Choose a box. The project is inside.","Choisis une cassette sur le rayon.":"Choose a tape from the shelf.","ChromeHue est une extension Chrome de contrôle des ampoules Philips Hue, publiée par MarlBurroW en 2015. Sa présentation historique et ses captures sont encore accessibles dans l'annonce de l'auteur.":"ChromeHue is a Chrome extension for controlling Philips Hue bulbs, published by MarlBurroW in 2015. Its original presentation and screenshots are still available in the author's announcement.","Chronologie":"Timeline","Chronologie officielle du Grand Challenge, DARPA":"Official Grand Challenge timeline, DARPA","classements d'outils IA":"AI tool rankings","Claude 3 introduit Haiku, Sonnet et Opus":"Claude 3 introduces Haiku, Sonnet and Opus","Claude 3.5 Sonnet devient disponible, avec des améliorations en code, compréhension et vision.":"Claude 3.5 Sonnet becomes available, with improvements in coding, understanding and vision.","Claude 3.7 Sonnet et Claude Code arrivent ensemble":"Claude 3.7 Sonnet and Claude Code arrive together","Claude 4 introduit Opus 4 et Sonnet 4":"Claude 4 introduces Opus 4 and Sonnet 4","Claude Haiku 5.5 complète la famille Claude 5.5":"Claude Haiku 5.5 completes the Claude 5.5 family","Claude identifie un système enzymatique associé à des répétitions d'ADN":"Claude identifies an enzyme system associated with DNA repeats","Claude Sonnet 5.5 succède à Sonnet 5":"Claude Sonnet 5.5 succeeds Sonnet 5","ClawScope est un tableau de bord local pour OpenClaw. Il regroupe la recherche dans la mémoire, une chronologie et un graphe de connaissances.":"ClawScope is a local dashboard for OpenClaw. It combines memory search, a timeline and a knowledge graph.","CLIP apprend conjointement à partir d'images et de descriptions pour reconnaître des concepts visuels sans réentraînement pour chaque catégorie.":"CLIP jointly learns from images and descriptions to recognize visual concepts without retraining for each category.","CLIP relie les images et les mots":"CLIP connects images and words","Cliquer pour ramasser":"Click to pick up","Cliquez sur un point pour consulter l'annonce.":"Click a point to view the announcement.","Code source":"Source code","Code Studio est l'interface web du daemon Akasha pour les projets de développement. Elle permet de créer des espaces de travail, demander des modifications aux agents et éditer les fichiers.":"Code Studio is the Akasha daemon's web interface for development projects. It lets you create workspaces, request changes from agents and edit files.","Codex ouvre un agent de développement dans le cloud":"Codex opens a cloud development agent","Cognition présente Devin, un agent de développement doté d'un terminal, d'un éditeur et d'un navigateur.":"Cognition introduces Devin, a development agent with a terminal, editor and browser.","Cognition présente l'agent de développement Devin":"Cognition introduces the Devin development agent","Commission européenne : entrée en vigueur":"European Commission: entry into force","Communiqué de l'Académie royale des sciences":"Royal Academy of Sciences press release","Communiqué officiel de Jeopardy!":"Official Jeopardy! announcement","Comparer les modèles":"Compare models","Comparer ses opinions aux votes et programmes.":"Compare your opinions with votes and manifestos.","Confidentialité":"Privacy","Confidentialité et durées de conservation":"Privacy and retention periods","Configurer les moniteurs":"Configure monitors","Connecter un assistant IA à TeamSpeak.":"Connect an AI assistant to TeamSpeak.","Consulter la chronologie":"View the timeline","Consultez vos moniteurs Uptime Kuma depuis Raycast. L'extension affiche leur disponibilité, leur temps de réponse et leurs derniers heartbeats.":"Check your Uptime Kuma monitors from Raycast. The extension shows their availability, response times and latest heartbeats.","Contrôle du plasma, DeepMind et EPFL":"Plasma control, DeepMind and EPFL","Contrôler les ampoules Philips Hue depuis Chrome.":"Control Philips Hue bulbs from Chrome.","Conversations, agents et outils en local.":"Conversations, agents and tools running locally.","Courbe chronologique. Flèches pour parcourir les événements, Début et Fin pour les bornes.":"Timeline chart. Use arrows to browse events, Home and End to reach the boundaries.","CUDA 1.0 ouvre le calcul sur GPU":"CUDA 1.0 opens GPU computing","culture":"culture","Cumul des événements retenus. Les dates et les sources figurent dans la liste chronologique.":"Cumulative selected events. Dates and sources appear in the chronological list.","Cursor Chat est une application multi-utilisateurs créée avec Cursor pour expérimenter le développement par conversations avec l'IA. Elle propose des échanges en temps réel et conserve l'historique des messages.":"Cursor Chat is a multiuser application built with Cursor to experiment with development through AI conversations. It supports real-time exchanges and keeps message history.","DALL-E 2 améliore la fidélité des images et introduit des fonctions de variation et de modification, avec un accès initial restreint.":"DALL-E 2 improves image fidelity and introduces variation and editing features, with initially restricted access.","DALL-E 2 entre en préversion limitée":"DALL-E 2 enters a limited preview","DALL-E génère des images à partir de texte":"DALL-E generates images from text","DALL-E traduit des descriptions en images et combine des concepts visuels dans une même génération.":"DALL-E turns descriptions into images and combines visual concepts in a single generation.","DALL·E 3 devient accessible dans ChatGPT":"DALL·E 3 becomes available in ChatGPT","DATA UNAVAILABLE":"DATA UNAVAILABLE","Date de publication invalide ou édition future.":"Invalid publication date or future edition.","Date documentée":"Documented date","Date historique invalide":"Invalid historical date","Datenschutz und Speicherdauer":"Privacy and retention periods","De":"From","Decisions API utilise Luna pour répondre à des questions dont les réponses possibles sont définies à l'avance.":"Decisions API uses Luna to answer questions with predefined possible answers.","Découvrir BallCam":"Discover BallCam","Découvrir la tirelire virtuelle":"Discover the virtual piggy bank","Découvrir le vidéoclub des projets":"Discover the project video club","Deep Blue gagne une partie contre Kasparov":"Deep Blue wins a game against Kasparov","Deep Blue remporte le match contre Kasparov":"Deep Blue wins the match against Kasparov","Deep Blue remporte une première partie contre le champion du monde, qui gagne ensuite le match.":"Deep Blue wins an initial game against the world champion, who then wins the match.","Deep Residual Learning for Image Recognition":"Deep Residual Learning for Image Recognition","Deep Review ouvre la relecture technique de documents":"Deep Review opens technical document review","DeepFace associe alignement 3D et réseau profond pour déterminer si deux photographies montrent la même personne.":"DeepFace combines 3D alignment and a deep network to determine whether two photographs show the same person.","DeepFace vérifie l'identité dans des images de visages":"DeepFace verifies identity in facial images","DeepMind annonce les résultats d'AlphaFold à CASP14 pour prédire la structure tridimensionnelle des protéines.":"DeepMind announces AlphaFold's CASP14 results for predicting three-dimensional protein structures.","DeepMind présente AlphaEvolve, un agent qui propose et améliore des algorithmes à l'aide de Gemini et d'évaluateurs automatiques.":"DeepMind introduces AlphaEvolve, an agent that proposes and improves algorithms using Gemini and automatic evaluators.","DeepMind publie AlphaGeometry, qui associe modèle neuronal et moteur de déduction pour résoudre des problèmes de géométrie.":"DeepMind releases AlphaGeometry, combining a neural model and a deduction engine to solve geometry problems.","DeepSeek ouvre V4-Pro et V4-Flash en préversion, sur le Web et en API, avec publication des poids.":"DeepSeek opens V4-Pro and V4-Flash in preview on the Web and through an API, with published weights.","DeepSeek publie R1 et ses poids sous licence MIT, avec un accès sur le Web et en API.":"DeepSeek releases R1 and its weights under the MIT license, with Web and API access.","DeepSeek publie V3, un modèle à mélange d'experts de 671 milliards de paramètres totaux, dont 37 milliards activés.":"DeepSeek releases V3, a mixture-of-experts model with 671 billion total parameters, of which 37 billion are active.","DeepSeek-R1 ouvre un modèle de raisonnement":"DeepSeek-R1 opens a reasoning model","DeepSeek-V3 publie ses poids":"DeepSeek-V3 releases its weights","DeepSeek-V4 Preview ouvre un contexte d'un million de tokens":"DeepSeek-V4 Preview opens a million-token context","Démonstration":"Demo","Démonstrations":"Demos","Denoising Diffusion Probabilistic Models":"Denoising Diffusion Probabilistic Models","Déploiement dans l'application Gemini":"Rollout in the Gemini app","Dépôt des manuscrits, corrections et formalisations":"Repository of manuscripts, corrections and formalizations","Des agents Claude repèrent un système de transcriptases inverses associé à des répétitions d'ADN, nommé ART, puis étudié au laboratoire.":"Claude agents identify a reverse-transcriptase system associated with DNA repeats, named ART, which is then studied in the laboratory.","Des histoires pour réviser du CP au CM2.":"Stories for reviewing the curriculum from CP to CM2.","Des modèles aux tâches":"From models to tasks","Des notifications vocales sur Google Home.":"Voice notifications on Google Home.","Des réseaux traduisent une séquence en une autre":"Networks translate one sequence into another","DES RETOURS":"FEEDBACK","Destination absente.":"Missing destination.","Deux réseaux apprennent en compétition : l'un produit des exemples et l'autre tente de reconnaître les exemples synthétiques.":"Two networks learn in competition: one produces examples and the other tries to identify synthetic examples.","DevDay 2026, Agents API":"DevDay 2026, Agents API","DevDay 2026, préversion Decisions API":"DevDay 2026, Decisions API preview","Disponibilité officielle dans Amazon Bedrock":"Official availability in Amazon Bedrock","Distribution Python":"Python distribution","Document de lancement et de sécurité GPT-6 Astra":"GPT-6 Astra launch and safety document","DQN apprend à jouer à Atari":"DQN learns to play Atari","Earth AI étend ses agents d'analyse géospatiale":"Earth AI expands its geospatial analysis agents","ÉDITION MONDIALE":"WORLD EDITION","Édition publique vérifiée.":"Public edition verified.","Édition source invalide.":"Invalid source edition.","education":"education","Éducation":"Education","Einwilligung widerrufen":"Withdraw consent","Emplacement absent.":"Missing location.","En ligne":"Online","En recherche, les fiches précisent la portée du résultat : preuve proposée, vérification formelle, expérience ou essai clinique.":"For research, entries specify the scope of the result: proposed proof, formal verification, experiment or clinical trial.","End":"End","Entrée en vigueur":"Entry into force","Entrer dans le vidéoclub":"Enter the video club","Envoyez une requête HTTP pour faire lire une notification par vos appareils Google Home. Le panneau d'administration configure les appareils, les clés API et les requêtes à envoyer depuis un script ou une automatisation IFTTT.":"Send an HTTP request to have your Google Home devices read a notification. The administration panel configures devices, API keys and requests sent from a script or IFTTT automation.","Épisode absent.":"Missing episode.","Essai clinique publié dans Nature Medicine":"Clinical trial published in Nature Medicine","Essaie un autre titre ou le nom d'un créateur.":"Try another title or a creator's name.","Étude GNoME et publication Nature":"GNoME study and Nature publication","Étude RFdiffusion dans Nature":"RFdiffusion study in Nature","Évaluation IMO 2024 par Google DeepMind":"IMO 2024 evaluation by Google DeepMind","Évaluation officielle IMO 2025":"Official IMO 2025 evaluation","Évaluer des choix et contrôler les outils.":"Evaluate choices and control tools.","événement":"event","Événement":"Event","Événement hors de la période vérifiée":"Event outside the verified period","Événement précédent":"Previous event","Événement suivant":"Next event","événements":"events","événements documentés.":"documented events.","Explorateur chronologique":"Timeline explorer","Explorer :":"Explore:","Explorer l'échelle d'une explosion en 3D.":"Explore the scale of an explosion in 3D.","Explorer l'histoire ou son récapitulatif annuel":"Explore history or its annual recap","Explorer la simulation":"Explore the simulation","Explorer les données politiques":"Explore political data","Export":"Export","Exporter les données":"Export data","Exposition et vérification par Terence Tao":"Explanation and verification by Terence Tao","Extension":"Extension","Extension de la base AlphaFold":"AlphaFold database expansion","Extraits sur Doom et les échecs":"Clips on Doom and chess","FAITES":"GIVE","Fiche de l'événement sélectionné":"Selected event details","Fiche officielle de Gemini 3.8 Flash":"Official Gemini 3.8 Flash model card","Fiche système officielle du 6 avril 2022":"Official system card dated April 6, 2022","Fiche vérifiée le":"Entry verified on","Filtrer par années":"Filter by year","Formulation du problème, Clay Mathematics Institute":"Problem statement, Clay Mathematics Institute","FunSearch améliore des constructions mathématiques":"FunSearch improves mathematical constructions","FunSearch combine un modèle de langage et un évaluateur de programmes pour trouver de nouvelles constructions pour les cap sets et des heuristiques de rangement.":"FunSearch combines a language model and a program evaluator to find new cap-set constructions and packing heuristics.","games":"games","Gardez l'état de vos services Uptime Kuma sous les yeux sur votre Stream Deck. Chaque touche affiche une mesure et peut changer l'information affichée ou suspendre un moniteur.":"Keep your Uptime Kuma service status in view on your Stream Deck. Each key displays a metric and can change the displayed information or pause a monitor.","Gemini 1.5 explore le million de tokens":"Gemini 1.5 explores a million tokens","Gemini 2.5 intègre le raisonnement":"Gemini 2.5 integrates reasoning","Gemini 3 arrive dans l'application et la recherche":"Gemini 3 arrives in the app and search","Gemini 3.1 Pro ouvre sa préversion":"Gemini 3.1 Pro opens its preview","Gemini 4 Argon arrive dans l'API et les applications Google":"Gemini 4 Argon arrives in the API and Google apps","Gemini Deep Think atteint le niveau or à l'IMO":"Gemini Deep Think reaches gold-medal level at the IMO","Gemini Robotics relie compréhension visuelle et action physique":"Gemini Robotics connects visual understanding and physical action","Generative Adversarial Networks":"Generative Adversarial Networks","Générer, éditer et mixer un morceau.":"Generate, edit and mix a track.","GitHub #1 Category":"GitHub #1 Category","GITHUB AI AWAITING DATA":"GITHUB AI AWAITING DATA","GITHUB AI DATA OUTDATED":"GITHUB AI DATA OUTDATED","GITHUB AI DATA UNAVAILABLE":"GITHUB AI DATA UNAVAILABLE","GITHUB AI NOT CURRENTLY RANKED":"GITHUB AI NOT CURRENTLY RANKED","GitHub Copilot entre dans l'éditeur de code":"GitHub Copilot enters the code editor","GLM-5.3 renforce le code et les tâches longues":"GLM-5.3 strengthens coding and long tasks","GNoME explore la stabilité de nouveaux matériaux":"GNoME explores the stability of new materials","GNoME propose des structures cristallines et évalue leur stabilité avec des calculs de physique des matériaux.":"GNoME proposes crystal structures and evaluates their stability using materials physics calculations.","Google annonce Gemini 1.5 Pro, avec une fenêtre expérimentale pouvant atteindre un million de tokens.":"Google announces Gemini 1.5 Pro, with an experimental window of up to one million tokens.","google deepmind":"google deepmind","Google déploie Gemini 3.1 Pro pour les tâches complexes de raisonnement.":"Google rolls out Gemini 3.1 Pro for complex reasoning tasks.","Google dévoile Gemini 1.0 en trois tailles : Ultra, Pro et Nano.":"Google unveils Gemini 1.0 in three sizes: Ultra, Pro and Nano.","Google lance Gemini 2.0 Flash Experimental":"Google launches Gemini 2.0 Flash Experimental","Google lance Gemini 2.0 Flash Experimental et présente de nouvelles capacités multimodales et d'utilisation d'outils.":"Google launches Gemini 2.0 Flash Experimental and introduces new multimodal and tool-use capabilities.","Google ouvre Deep Review, un système d'agents qui analyse les arguments et la cohérence de documents techniques et scientifiques.":"Google opens Deep Review, an agent system that analyzes the arguments and consistency of technical and scientific documents.","Google ouvre Gemini 2.5 Pro Experimental, premier modèle de sa nouvelle génération avec raisonnement intégré.":"Google opens Gemini 2.5 Pro Experimental, the first model in its new generation with built-in reasoning.","Google ouvre Gemini 3 avec de nouvelles capacités de raisonnement et des réponses sous forme d'interfaces générées.":"Google opens Gemini 3 with new reasoning capabilities and answers in the form of generated interfaces.","Google ouvre Gemini 4 Argon pour les tâches longues de développement, de raisonnement et de recherche.":"Google opens Gemini 4 Argon for long development, reasoning and research tasks.","Google ouvre une API de recherche pour AlphaGenome, qui prédit plusieurs propriétés de régulation des gènes à partir de séquences d'ADN.":"Google opens a research API for AlphaGenome, which predicts several gene regulation properties from DNA sequences.","Google présente Aletheia, un agent basé sur Deep Think qui génère, vérifie et révise des arguments mathématiques.":"Google introduces Aletheia, a Deep Think-based agent that generates, verifies and revises mathematical arguments.","Google présente Gemini Robotics et Robotics-ER, deux modèles dérivés de Gemini 2.0 pour la manipulation et le raisonnement spatial.":"Google introduces Gemini Robotics and Robotics-ER, two models derived from Gemini 2.0 for manipulation and spatial reasoning.","Google présente la famille Gemini":"Google introduces the Gemini family","Google présente LaMDA pour la conversation":"Google introduces LaMDA for conversation","Google présente PaLM, entraîné avec Pathways, et évalue ses capacités sur des tâches de langage, de raisonnement et de code.":"Google introduces PaLM, trained with Pathways, and evaluates its language, reasoning and coding capabilities.","Google présente un agent qui combine modèles spécialisés, données géospatiales et outils pour analyser des phénomènes à l'échelle de la planète.":"Google introduces an agent combining specialized models, geospatial data and tools to analyze phenomena at planetary scale.","Google présente un assistant de recherche scientifique multi-agent":"Google introduces a multiagent scientific research assistant","Google présente Veo 3, capable de générer des vidéos avec dialogues, ambiances et effets sonores.":"Google introduces Veo 3, capable of generating videos with dialogue, ambience and sound effects.","Google publie la fiche de Gemini 3.8 Flash":"Google publishes the Gemini 3.8 Flash model card","Google publie TensorFlow sous licence Apache 2.0 pour entraîner et exécuter des modèles d'apprentissage automatique.":"Google releases TensorFlow under Apache 2.0 to train and run machine learning models.","governance":"governance","GPT-2 est annoncé avec une diffusion progressive":"GPT-2 is announced with a phased release","GPT-3 apprend à partir de quelques exemples":"GPT-3 learns from a few examples","GPT-3 devient accessible par une API privée":"GPT-3 becomes accessible through a private API","GPT-4 arrive dans ChatGPT et en API":"GPT-4 arrives in ChatGPT and the API","GPT-4o rapproche texte, vision et voix":"GPT-4o brings text, vision and voice together","GPT-5 réunit réponse rapide et raisonnement":"GPT-5 combines fast answers and reasoning","GPT-5.4 intègre l'utilisation native de l'ordinateur":"GPT-5.4 integrates native computer use","GPT-5.5 arrive dans ChatGPT et Codex":"GPT-5.5 arrives in ChatGPT and Codex","GPT-5.6 Sol arrive dans ChatGPT":"GPT-5.6 Sol arrives in ChatGPT","GPT-6 déploie les interfaces interactives dans ChatGPT":"GPT-6 rolls out interactive interfaces in ChatGPT","GPT-6 Sol et Luna complètent Astra":"GPT-6 Sol and Luna complement Astra","GPT-6.1 Sol améliore le code, l'utilisation de l'ordinateur et les tâches professionnelles par rapport à GPT-6 Sol.":"GPT-6.1 Sol improves coding, computer use and professional tasks compared with GPT-6 Sol.","GPT-6.1 Sol rapproche les capacités d'Astra à moindre coût":"GPT-6.1 Sol approaches Astra's capabilities at lower cost","Grandir à travers les époques de l'Histoire.":"Grow through the eras of history.","GraphCast produit des prévisions météo à dix jours":"GraphCast produces ten-day weather forecasts","GraphCast utilise un réseau de neurones sur graphes pour prévoir la météo mondiale jusqu'à dix jours à l'avance.":"GraphCast uses a graph neural network to forecast global weather up to ten days ahead.","Grok 4.7 renouvelle la gamme Grok":"Grok 4.7 updates the Grok range","Grok Bot ouvre sa bêta d'agents persistants":"Grok Bot opens its persistent-agent beta","Grok Bot propose des agents dotés d'un ordinateur dans le cloud, d'une mémoire et de routines pour travailler dans les outils de l'utilisateur.":"Grok Bot offers agents with a cloud computer, memory and routines to work in the user's tools.","Grok Voice 3.0 arrive dans l'API vocale":"Grok Voice 3.0 arrives in the voice API","Grok Voice 3.0 renouvelle le modèle de conversation vocale de xAI, avec gestion des tours de parole et des interruptions.":"Grok Voice 3.0 updates xAI's voice conversation model, with turn-taking and interruption handling.","H1118\" stroke=\"#203746\"/>":"H1118\" stroke=\"#203746\"/>","Héros de la classe propose des histoires interactives pour réviser le programme du CP au CM2. L'enfant choisit une aventure et prend des décisions au fil des pages. Ses choix orientent la suite du récit.":"Héros de la classe offers interactive stories to review the curriculum from CP to CM2. Children choose an adventure and make decisions as they read. Their choices shape the rest of the story.","High-Resolution Image Synthesis with Latent Diffusion Models":"High-Resolution Image Synthesis with Latent Diffusion Models","Hinton, Osindero et Teh présentent un apprentissage couche par couche pour les réseaux de croyance profonds.":"Hinton, Osindero and Teh present layer-by-layer learning for deep belief networks.","HiRO-ACEv2 simule les événements météo extrêmes à 3 km":"HiRO-ACEv2 simulates extreme weather events at 3 km","Histoire de Deep Blue, IBM":"History of Deep Blue, IBM","Histoire de l'intelligence artificielle":"History of artificial intelligence","Hivekeep est une plateforme à héberger chez soi pour faire travailler une équipe d'agents IA. Chaque agent conserve sa mémoire, sa conversation, ses outils et son espace de travail. Les agents peuvent se déléguer des tâches et collaborer.":"Hivekeep is a self-hosted platform for running a team of AI agents. Each agent keeps its memory, conversation, tools and workspace. Agents can delegate tasks to one another and collaborate.","Hochreiter et Schmidhuber proposent une architecture récurrente capable d'apprendre des dépendances temporelles longues.":"Hochreiter and Schmidhuber propose a recurrent architecture capable of learning long-term temporal dependencies.","Home":"Home","Icône de Aigora":"Aigora icon","Icône de Akasha":"Akasha icon","Icône de Akasha Code Studio":"Akasha Code Studio icon","Icône de Akasha OS":"Akasha OS icon","Icône de BallCam.tv":"BallCam.tv icon","Icône de Boostback":"Boostback icon","Icône de Bot Chronicles":"Bot Chronicles icon","Icône de Héros de la classe":"Héros de la classe icon","Icône de Hivekeep":"Hivekeep icon","Icône de MindOdyssey":"MindOdyssey icon","Icône de Mission Maison":"Mission Maison icon","Icône de Not My Fault":"Not My Fault icon","Icône de Okodukai":"Okodukai icon","Icône de PinchChat":"PinchChat icon","Icône de Point Zero":"Point Zero icon","Icône de Politiquizz":"Politiquizz icon","Icône de Raycast Uptime Kuma":"Raycast Uptime Kuma icon","Icône de Rbitnet":"Rbitnet icon","Icône de Sillage":"Sillage icon","Icône de Song Maker":"Song Maker icon","Icône de Streamdeck Uptime Kuma":"Streamdeck Uptime Kuma icon","Identifiant dupliqué":"Duplicate identifier","Ihre Auswahl konnte nicht gespeichert werden. Erlauben Sie den lokalen Speicher und versuchen Sie es erneut.":"Your choice could not be saved. Allow local storage and try again.","Il faut gérer la poussée et la trajectoire pendant le vol retour, puis réussir l'approche de la tour. Le jeu se lance gratuitement dans un navigateur sur ordinateur et propose un classement des pilotes.":"You must manage thrust and trajectory on the return flight, then complete the approach to the tower. The game runs free in a desktop browser and offers a pilot leaderboard.","Il peut envoyer des alertes sur des mots-clés, se reconnecter et gérer plusieurs serveurs. Un système de plugins étend son comportement. Ce projet ancien est présenté à partir de son code public et de sa documentation.":"It can send keyword alerts, reconnect and manage multiple servers. Plugins extend its behavior. This older project is presented using its public code and documentation.","Il prend en charge plusieurs types d'enregistrements, la gestion de l'IP publique et une détection optionnelle des labels Traefik. Le README présente le projet comme un travail en cours.":"It supports several record types, public IP management and optional detection of Traefik labels. The README describes the project as work in progress.","Il sert notamment de moteur local à Akasha. Le dépôt publie les versions et les instructions pour l'exécuter ou l'intégrer à d'autres outils.":"It notably serves as Akasha's local engine. The repository publishes releases and instructions for running it or integrating it with other tools.","Image requise":"Image required","Imagen associe langage et diffusion":"Imagen combines language and diffusion","Imagen combine un grand encodeur de langage avec des modèles de diffusion pour générer des images à partir de descriptions.":"Imagen combines a large language encoder with diffusion models to generate images from descriptions.","ImageNet présente une base d'images annotées et organisées selon une hiérarchie destinée à la recherche en vision.":"ImageNet introduces an annotated image database organized in a hierarchy for vision research.","ImageNet rassemble 3,2 millions d'images annotées":"ImageNet gathers 3.2 million annotated images","Index de recherche OpenAI, annonce du 29 septembre":"OpenAI research index, September 29 announcement","Informations":"Information","InstructGPT suit les consignes humaines":"InstructGPT follows human instructions","Introducing OpenAI, annonce originale":"Introducing OpenAI, original announcement","jalons à retenir":"key milestones","JALONS DOCUMENTÉS / CUMUL":"DOCUMENTED MILESTONES / CUMULATIVE","Jalons majeurs uniquement":"Major milestones only","Jeux":"Games","John Hopfield et Geoffrey Hinton reçoivent le Nobel de physique pour leurs travaux fondateurs sur les réseaux de neurones.":"John Hopfield and Geoffrey Hinton receive the Nobel Prize in Physics for their foundational work on neural networks.","Journal de publication officiel de Wan2.1":"Official Wan2.1 release log","Journal officiel de publication SDXL":"Official SDXL release log","Kingma et Welling décrivent une méthode d'apprentissage de modèles génératifs à variables latentes, à l'origine des VAE.":"Kingma and Welling describe a learning method for latent-variable generative models, the foundation of VAEs.","Kingma et Welling présentent les autoencodeurs variationnels":"Kingma and Welling introduce variational autoencoders","L":"L","L'accueil de l'enfant":"The child's home screen","L'AI Act entre en vigueur en Europe":"The AI Act enters into force in Europe","L'année en IA":"The year in AI","L'application réunit les outils de production et de mix, une partition MIDI/ABC et les exports WAV, FLAC et MP3. Le code et les versions pour Windows, macOS et Linux sont publics.":"The application brings together production and mixing tools, a MIDI/ABC score and WAV, FLAC and MP3 exports. The code and versions for Windows, macOS and Linux are public.","L'apprentissage profond identifie l'halicine comme antibiotique":"Deep learning identifies halicin as an antibiotic","L'arrangement du morceau":"The track arrangement","L'article sur GPT-3 étudie un modèle de 175 milliards de paramètres capable d'effectuer des tâches à partir d'exemples dans le prompt.":"The GPT-3 paper studies a 175-billion-parameter model capable of performing tasks from examples in the prompt.","L'écran affiche un avatar pendant les échanges. Le dépôt fournit les guides de flash, d'utilisation et de connexion au daemon. Il s'agit d'un projet matériel à monter, pas d'un appareil commercial prêt à l'emploi.":"The screen shows an avatar during conversations. The repository provides flashing, usage and daemon connection guides. This is a hardware project to assemble, not a ready-to-use commercial device.","L'édition des clips":"Clip editing","l'équipe Sandbox":"the Sandbox team","L'équipe Wan publie le code d'inférence et les poids de Wan2.1.":"The Wan team publishes the inference code and weights for Wan2.1.","L'espace de fichiers":"The file area","L'état de vos services sur votre Stream Deck.":"Your services' status on your Stream Deck.","L'étude montre comment répartir le budget de calcul entre taille du modèle et quantité de données d'entraînement.":"The study shows how to distribute the compute budget between model size and the amount of training data.","L'extraction de faits et le graphe de connaissances relient les informations. Un plugin pour OpenClaw, un serveur MCP et une interface en ligne de commande sont publiés. Le projet est développé sous le compte AkashaBot, piloté par Loïc.":"Fact extraction and the knowledge graph link information together. An OpenClaw plugin, MCP server and command-line interface are published. The project is developed under the AkashaBot account, run by Loïc.","L'IA générative":"Generative AI","L'interface web regroupe les conversations, fichiers, mini-apps et tâches programmées. Plusieurs messageries peuvent être reliées aux agents, dont Telegram, WhatsApp, Slack, Discord, Signal et Matrix. Hivekeep est le nouveau nom de KinBot.":"The web interface brings together conversations, files, mini-apps and scheduled tasks. Agents can connect to messaging services including Telegram, WhatsApp, Slack, Discord, Signal and Matrix. Hivekeep is the new name for KinBot.","L'univers de Bot Chronicles":"The world of Bot Chronicles","L'univers de MindOdyssey":"The world of MindOdyssey","L'univers de Point Zero":"The world of Point Zero","L'univers de Rbitnet":"The world of Rbitnet","La base AlphaFold étend son accès libre aux structures prédites de presque toutes les protéines cataloguées par la science.":"The AlphaFold database extends free access to predicted structures of nearly all proteins catalogued by science.","La carte politique des modèles de langage.":"The political map of language models.","La conversation avec l'assistant":"The conversation with the assistant","La diffusion génère des images par débruitage":"Diffusion generates images by denoising","La diffusion passe dans un espace latent":"Diffusion moves into latent space","La disponibilité actuelle de l'extension n'a pas été confirmée. Cette fiche conserve la trace du projet et renvoie à sa présentation d'origine.":"The extension's current availability has not been confirmed. This entry preserves a record of the project and links to its original presentation.","La fiche dans la boîte":"The project card inside the box","La fiche de Gemini 3.8 Flash décrit des progrès en ingénierie logicielle et dans les tâches professionnelles réalisées par des agents.":"The Gemini 3.8 Flash model card describes progress in software engineering and professional tasks performed by agents.","La liste des moniteurs":"The monitor list","La mission Ship Happens":"The Ship Happens mission","La preuve de cette édition ne permet pas de vérifier son profil de calcul. Aucun autre profil ne lui est attribué.":"This edition's proof does not allow its calculation profile to be verified. No other profile is assigned to it.","La préversion technique de Copilot propose des lignes et des fonctions à partir du code environnant grâce à OpenAI Codex.":"The Copilot technical preview suggests lines and functions from surrounding code using OpenAI Codex.","LA SÉLECTION SANDBOX":"THE SANDBOX SELECTION","La version Preview comprend la gestion des agents, des modèles et des espaces de création. Elle s'installe sur un système existant : le projet ne distribue pas encore une image de système d'exploitation amorçable.":"The Preview version includes management of agents, models and creative workspaces. It installs on an existing system: the project does not yet distribute a bootable operating system image.","La vue scindée, la recherche et l'export Markdown servent à retrouver les échanges. Les appels d'outils et la consommation de contexte restent visibles. L'interface existe en français et en anglais et peut s'installer comme une PWA.":"Split view, search and Markdown export help retrieve conversations. Tool calls and context usage remain visible. The interface is available in French and English and can be installed as a PWA.","LaMDA est présenté comme un modèle entraîné sur le dialogue pour suivre des conversations ouvertes sur des sujets variés.":"LaMDA is introduced as a dialogue-trained model for open-ended conversations on various topics.","Lancement de Gemini 3 dans l'application":"Gemini 3 launch in the app","Lancement de GPT-6 avec Intelligent UI":"GPT-6 launch with Intelligent UI","Lancement public de Sora":"Public launch of Sora","Language Models are Few-Shot Learners":"Language Models are Few-Shot Learners","Le calculateur d'IBM bat Garry Kasparov sur un match de six parties aux conditions de tournoi.":"IBM's computer defeats Garry Kasparov in a six-game match under tournament conditions.","Le carnet de recettes":"The recipe book","Le chat relié au serveur Rust":"The chat connected to the Rust server","Le choix du lieu et de la puissance modifie la simulation. Les estimations d'impact humain utilisent les données de population WorldPop. Elles sont illustratives et ne constituent pas une prévision opérationnelle.":"The location and yield choices change the simulation. Human impact estimates use WorldPop population data. They are illustrative and do not constitute an operational forecast.","Le ciel depuis trois observateurs":"The sky from three observers","Le classement public est en retard.":"The public ranking is overdue.","Le code documente le suivi des coûts et de la latence ainsi que la désactivation temporaire des modèles en échec. Le projet est publié sous le compte AkashaBot, piloté par Loïc.":"The code documents cost and latency tracking as well as temporary disabling of failing models. The project is published under the AkashaBot account, run by Loïc.","Le Coffre magique":"The magic chest","LE COIN":"THE CORNER","LE COIN DES CURIEUX":"THE CURIOUS CORNER","Le deep learning":"Deep learning","Le détail du moniteur":"Monitor details","Le détail du replay":"Replay details","Le langage change d'échelle":"Language scales up","Le même ciel, vu depuis trois endroits.":"The same sky, seen from three places.","Le mode agent de ChatGPT intègre les capacités d'Operator pour réaliser des tâches dans un navigateur.":"ChatGPT's agent mode integrates Operator's capabilities to perform tasks in a browser.","Le Nobel de chimie distingue la prédiction des protéines":"The Nobel Prize in Chemistry recognizes protein prediction","Le Nobel de chimie récompense David Baker, Demis Hassabis et John Jumper.":"The Nobel Prize in Chemistry honors David Baker, Demis Hassabis and John Jumper.","Le Nobel de physique récompense Hopfield et Hinton":"The Nobel Prize in Physics honors Hopfield and Hinton","Le panneau de commandement":"The command panel","Le panneau de navigation":"The navigation panel","Le panneau de notifications":"The notifications panel","Le plugin suit l'état, les temps de réponse et la disponibilité sur 24 heures ou 30 jours. Les mises à jour passent par Socket.IO. Le manifeste indique Windows et macOS, avec une prise en charge du Stream Deck+.":"The plugin tracks status, response times and availability over 24 hours or 30 days. Updates use Socket.IO. The manifest lists Windows and macOS, with Stream Deck+ support.","Le podcast IA et tech de Lou, Loïc et Nico. Chaque semaine, leurs essais d'outils IA et les projets qu'ils construisent avec.":"Lou, Loïc and Nico's AI and tech podcast. Every week, their AI tool experiments and the projects they build with them.","Le premier GPT est présenté":"The first GPT is introduced","Le quiz des votes":"The voting quiz","Le règlement européen sur l'intelligence artificielle entre en vigueur.":"The European Union's artificial intelligence regulation enters into force.","Le relais WebSocket gère les messages, la présence, l'authentification et la reconnexion. Le service et son plugin OpenClaw sont publiés dans le même dépôt.":"The WebSocket relay handles messages, presence, authentication and reconnection. The service and its OpenClaw plugin are published in the same repository.","Le retrait exige --id ou --week, sans filtre.":"Withdrawal requires --id or --week, without a filter.","Le retrait ne prend pas --dry-run. Aucune action exécutée.":"Withdrawal does not accept --dry-run. No action executed.","Le salon des agents":"The agent room","Le site permet aussi de comparer les candidats, de consulter les lois et les votes, de parcourir les thèmes de société et d'examiner les engagements des quinquennats. Les informations sont reliées à leurs sources publiques et une page explique la méthode.":"The site also lets users compare candidates, consult laws and votes, browse social issues and examine commitments from five-year presidential terms. Information links to public sources and a page explains the method.","Le site rassemble leurs publications et illustrations, une présentation des auteurs et un flux RSS. Le code des pages et des articles est public.":"The site gathers their posts and illustrations, an introduction to the authors and an RSS feed. The code for the pages and articles is public.","Le sitemap ne contient aucune URL.":"The sitemap contains no URLs.","Le Stream Deck en service":"The Stream Deck in use","Le suivi de la consommation":"Usage tracking","Le tableau de bord des parents":"The parent dashboard","Le tableau de bord suit aussi les sessions, activités et tâches planifiées. Il utilise la base SQLite du projet Offline Memory. Le code est publié sous le compte AkashaBot, piloté par Loïc.":"The dashboard also tracks sessions, activities and scheduled tasks. It uses the Offline Memory project's SQLite database. The code is published under the AkashaBot account, run by Loïc.","Le téléchargement a échoué. Réessayez.":"The download failed. Try again.","Le Transformer traite les séquences par attention":"The Transformer processes sequences through attention","Le véhicule autonome de Stanford termine en tête une course de 132 miles dans le désert, sans conducteur.":"Stanford's autonomous vehicle finishes first in a 132-mile desert race without a driver.","Le vidéoclub":"The video club","Le village préhistorique":"The prehistoric village","Le vol et les instruments":"The flight and instruments","Le voyage attend un instant.":"The journey needs a moment.","LeCun et ses coauteurs décrivent des réseaux convolutifs appliqués à la reconnaissance de documents et de chèques.":"LeCun and coauthors describe convolutional networks applied to document and check recognition.","LeNet-5 lit les chiffres manuscrits":"LeNet-5 reads handwritten digits","Les actions servent à ouvrir un moniteur ou son URL, à le suspendre et à le reprendre. L'extension accepte l'authentification à deux facteurs et suit les changements avec Socket.IO.":"Actions open a monitor or its URL, pause it and resume it. The extension supports two-factor authentication and tracks changes using Socket.IO.","Les applications, jeux et outils créés par l'équipe Sandbox. Découvrez leurs captures, leurs vidéos et les projets open source.":"Apps, games and tools created by the Sandbox team. Discover their screenshots, videos and open-source projects.","Les archives n'ont pas pu être chargées.":"The archives could not be loaded.","Les aventures de Héros de la classe":"The adventures of Héros de la classe","Les chroniques de deux agents IA.":"The chronicles of two AI agents.","Les connexions résiduelles permettent d'entraîner des réseaux beaucoup plus profonds pour la reconnaissance d'images.":"Residual connections allow much deeper networks to be trained for image recognition.","Les conversations":"Conversations","Les conversations et le briefing":"Conversations and briefing","Les conversations et les outils":"Conversations and tools","Les créations de Lou, Loïc et Nicolas, en VHS.":"Lou, Loïc and Nicolas's creations, on VHS.","Les DDPM apprennent à reconstruire progressivement des images à partir de bruit.":"DDPMs learn to progressively reconstruct images from noise.","Les deux systèmes résolvent quatre des six problèmes des Olympiades internationales de mathématiques 2024, soit 28 points sur 42.":"The two systems solve four of the six problems at the 2024 International Mathematical Olympiad, scoring 28 points out of 42.","Les Dots utilisent GPT-6 Astra et leur propre ordinateur dans le cloud pour poursuivre des objectifs, conserver du contexte et utiliser des applications connectées.":"Dots use GPT-6 Astra and their own cloud computer to pursue goals, retain context and use connected apps.","Les espaces de création":"Creative workspaces","Les fiches reprennent les sites, dépôts et publications publiques des créateurs. Les sources et la date de vérification figurent sur chaque fiche.":"Entries draw on creators' public websites, repositories and posts. Sources and the verification date appear on each entry.","Les fichiers à côté de la conversation":"Files alongside the conversation","Les fondations":"The foundations","Les GAN opposent générateur et discriminateur":"GANs pit a generator against a discriminator","Les grandes dates de l'IA":"AI's key dates","Les graphiques radar, classements et comparaisons rendent les résultats consultables. Le site annoncé renvoyait une erreur 404 lors de la vérification du 13 octobre 2026 ; la fiche renvoie donc au dépôt public et à sa méthode.":"Radar charts, rankings and comparisons make the results accessible. The announced site returned a 404 error during the October 13, 2026 check, so the entry links to the public repository and its method.","Les histoires associent illustrations, lecture et notions scolaires. La personnalisation du prénom permet à l'enfant de se retrouver dans l'aventure. Les profils, la progression sauvegardée et les cartes à collectionner accompagnent les prochaines visites.":"The stories combine illustrations, reading and school concepts. A personalized first name lets children see themselves in the adventure. Profiles, saved progress and collectible cards support later visits.","Les matières et leurs aventures":"Subjects and their adventures","Les mini-apps dans la conversation":"Mini-apps in the conversation","Les missions et le mode PvP":"Missions and PvP mode","Les modèles d'IA classés selon les mesures de chaque édition.":"AI models ranked using each edition's measurements.","Les modèles de diffusion latente réduisent les coûts de génération en travaillant sur une représentation compressée des images.":"Latent diffusion models reduce generation costs by working on a compressed representation of images.","Les moniteurs sur le Stream Deck":"Monitors on the Stream Deck","Les outils documentés couvrent aussi les permissions, groupes, diagnostics, logs, fichiers et snapshots. Le projet se distribue sur PyPI et peut fonctionner avec Python ou Docker.":"The documented tools also cover permissions, groups, diagnostics, logs, files and snapshots. The project is distributed through PyPI and can run with Python or Docker.","Les parents définissent les quêtes, valident leur réalisation et choisissent les récompenses. Ils règlent les retraits et l'accès aux placements. Les pièces ne s'achètent pas et ne se convertissent pas en euros. Les cartes à collectionner se gagnent en accomplissant les quêtes.":"Parents define quests, approve completion and choose rewards. They control withdrawals and access to investments. Coins cannot be purchased or converted into euros. Collectible cards are earned by completing quests.","Les prénoms, tâches et règles se modifient dans les réglages. Les données sont conservées dans le navigateur avec localStorage.":"First names, tasks and rules can be changed in settings. Data is kept in the browser using localStorage.","Les profils GitHub, LinkedIn et X des animateurs Sandbox.":"The Sandbox hosts' GitHub, LinkedIn and X profiles.","LES PROJETS DE":"PROJECTS BY","LES PROJETS DE L'ÉQUIPE":"THE TEAM'S PROJECTS","Les projets de Lou, Loïc et Nicolas dans un vidéoclub.":"Lou, Loïc and Nicolas's projects in a video club.","Les projets open source, d'après les mesures de chaque édition.":"Open-source projects, based on each edition's measurements.","LES PROJETS, COMME AU VIDÉOCLUB":"PROJECTS, VIDEO CLUB STYLE","Les quêtes à accomplir":"Quests to complete","Les résultats annoncés concernent notamment les empilements de sphères, les codes, les groupes et la complexité des calculs.":"The announced results cover areas including sphere packings, codes, groups and computational complexity.","Les séances de visionnage permettent de regarder ensemble et de discuter d'une décision de jeu. Le studio de clips sert à enregistrer un mouvement de caméra ou à définir un trajet avec des images clés, puis à partager le résultat. La bibliothèque publique se consulte sans compte.":"Viewing sessions let people watch together and discuss a game decision. The clip studio records camera movement or defines a path using keyframes, then shares the result. The public library requires no account.","Les skills et plugins étendent ses fonctions. Des canaux comme Telegram, Discord et Slack peuvent être reliés au même service. Le modèle embarqué fonctionne sans compte cloud obligatoire ; les fournisseurs externes restent optionnels.":"Skills and plugins extend its features. Channels such as Telegram, Discord and Slack can connect to the same service. The embedded model works without a mandatory cloud account; external providers remain optional.","Les skills, agents et MCP, avec leurs mesures et leurs sources.":"Skills, agents and MCP, with their measurements and sources.","Les tâches de la semaine":"The week's tasks","Les tâches de la semaine et les récompenses.":"The week's tasks and rewards.","Les tâches, aperçus, branches Git et builds sont documentés dans le guide d'utilisation. Le code est public ; aucun paquet npm ni version installable n'a été confirmé lors de la vérification.":"Tasks, previews, Git branches and builds are documented in the user guide. The code is public; no npm package or installable release was confirmed during verification.","Les Transformers":"Transformers","Les trois observateurs peuvent être déplacés. Les vues du ciel et à la première personne, l'orientation synchronisée et les réglages du firmament servent à examiner les différences entre les représentations.":"The three observers can be moved. Sky and first-person views, synchronized orientation and firmament settings help examine differences between representations.","Les utilisateurs connectés sont visibles et l'historique se charge progressivement. Le dépôt documente des thèmes clair et sombre et un déploiement Docker. Aucun site public actif n'a été confirmé.":"Connected users are visible and history loads progressively. The repository documents light and dark themes and Docker deployment. No active public site was confirmed.","Lien public entre Loïc et AkashaBot":"Public link between Loïc and AkashaBot","Lien Spotify à venir":"Spotify link coming soon","Lire la vidéo":"Play video","Livre blanc initial de TensorFlow":"Initial TensorFlow white paper","Llama 2 ouvre l'usage commercial":"Llama 2 opens commercial use","Llama 3 renouvelle les modèles accessibles":"Llama 3 updates the accessible models","Llama 3.1 publie un modèle de 405 milliards de paramètres":"Llama 3.1 releases a 405-billion-parameter model","Llama 4 introduit Scout et Maverick":"Llama 4 introduces Scout and Maverick","LLaMA ouvre une nouvelle famille de modèles":"LLaMA opens a new family of models","LLM Router expose un point d'entrée compatible avec OpenAI et sélectionne un modèle selon la tâche. Il peut associer OpenRouter, OpenAI, Anthropic, Google et Ollama dans des chaînes de repli.":"LLM Router exposes an OpenAI-compatible endpoint and selects a model for each task. It can combine OpenRouter, OpenAI, Anthropic, Google and Ollama in fallback chains.","Loïc":"Loïc","Lou":"Lou","LSTM donne de la mémoire aux réseaux":"LSTM gives networks memory","Markdown":"Markdown","MCP ouvre un protocole de connexion pour les assistants":"MCP opens a connection protocol for assistants","Measurement preferences":"Measurement preferences","Medición de audiencia":"Audience measurement","Menu":"Menu","MERCI DE":"THANK YOU FOR","Mesure d’audience":"Audience measurement","Meta lance Muse Wonder et Muse Wonder Full":"Meta launches Muse Wonder and Muse Wonder Full","Meta présente deux modèles pour les agents personnels et professionnels. Muse Wonder arrive dans Meta Model API et Muse Code.":"Meta introduces two models for personal and professional agents. Muse Wonder arrives in Meta Model API and Muse Code.","Meta présente Muse comme un agent personnel capable de suivre des objectifs et d'utiliser des services connectés.":"Meta introduces Muse as a personal agent capable of pursuing goals and using connected services.","Meta présente Muse, son agent personnel":"Meta introduces Muse, its personal agent","Meta publie Llama 3 en versions de 8 et 70 milliards de paramètres.":"Meta releases Llama 3 in 8- and 70-billion-parameter versions.","Meta publie Llama 3.1, dont une version à 405 milliards de paramètres et des modèles 8B et 70B actualisés.":"Meta releases Llama 3.1, including a 405-billion-parameter version and updated 8B and 70B models.","Meta publie Llama 4 Scout et Maverick, deux modèles nativement multimodaux à mélange d'experts.":"Meta releases Llama 4 Scout and Maverick, two natively multimodal mixture-of-experts models.","Meta publie LLaMA, une famille de modèles de langage de 7 à 65 milliards de paramètres.":"Meta releases LLaMA, a family of language models with 7 to 65 billion parameters.","Meta rend disponibles les modèles Llama 2 et Llama 2-Chat, pour la recherche et des usages commerciaux.":"Meta makes Llama 2 and Llama 2-Chat available for research and commercial use.","Méthode, dates, sources et limites de la version réellement publiée.":"Method, dates, sources and limits of the version actually published.","Mettre la lecture en pause":"Pause playback","Mikolov et ses coauteurs proposent des architectures efficaces pour apprendre des relations entre mots dans de grands corpus.":"Mikolov and coauthors propose efficient architectures for learning relationships between words in large corpora.","milestone":"milestone","MindOdyssey prépare un jeu d'aventure éducatif pour les enfants de 7 à 12 ans. L'enfant commence dans un village préhistorique. Les personnages lui confient des missions pour progresser et débloquer de nouveaux villages jusqu'à l'époque contemporaine.":"MindOdyssey is developing an educational adventure game for children aged 7 to 12. Children begin in a prehistoric village. Characters give them missions to progress and unlock new villages through to modern times.","Mise à jour antérieure à la publication.":"Update predates publication.","Mise à jour du 17 juillet sur l'intégration d'Operator":"July 17 update on Operator integration","Mission Maison suit les tâches familiales dans une interface adaptée au mobile. Les validations sont organisées par semaine, avec des points et des récompenses.":"Mission Maison tracks family tasks in a mobile-friendly interface. Approvals are organized by week, with points and rewards.","Mistral 3 publie une nouvelle génération ouverte":"Mistral 3 releases a new open generation","Mistral 7B, un modèle européen à poids ouverts":"Mistral 7B, a European open-weight model","mistral ai":"mistral ai","Mistral AI publie Mistral 7B, un modèle de langage de 7,3 milliards de paramètres sous licence Apache 2.0.":"Mistral AI releases Mistral 7B, a 7.3-billion-parameter language model under Apache 2.0.","Mistral Large 4 ouvre sa préversion publique":"Mistral Large 4 opens its public preview","Mistral ouvre l'API de Large 4, un modèle multimodal à mélange d'experts, pour le code et les tâches professionnelles.":"Mistral opens the API for Large 4, a multimodal mixture-of-experts model for coding and professional tasks.","Mistral publie Large 3 et trois petits modèles Ministral sous licence Apache 2.0.":"Mistral releases Large 3 and three small Ministral models under Apache 2.0.","Mistral publie Mixtral 8×7B, un modèle à mélange d'experts dont les poids sont disponibles sous Apache 2.0.":"Mistral releases Mixtral 8×7B, a mixture-of-experts model with weights available under Apache 2.0.","Mixtral diffuse l'architecture à experts":"Mixtral spreads the expert architecture","model":"model","Modèles, recherche et annonces mondiales.":"Models, research and worldwide announcements.","moonshot ai":"moonshot ai","music":"music","Musique":"Music","Nature Medicine publie un essai randomisé du rentosertib, inhibiteur de TNIK conçu avec l'IA, dans la fibrose pulmonaire idiopathique.":"Nature Medicine publishes a randomized trial of rentosertib, an AI-designed TNIK inhibitor, in idiopathic pulmonary fibrosis.","Nico":"Nico","Nicolas":"Nicolas","NO AWARD":"NO AWARD","No se pudo guardar tu elección. Permite el almacenamiento local e inténtalo de nuevo.":"Your choice could not be saved. Allow local storage and try again.","Non renseignée":"Not provided","Not My Fault est un jeu de communication : chaque joueur possède des outils et des informations que les autres ne voient pas. L'équipe doit décrire ses écrans, partager ses indices et décider avant la fin du compte à rebours.":"Not My Fault is a communication game: each player has tools and information the others cannot see. The team must describe their screens, share clues and decide before the countdown ends.","Notes officielles des modèles, 9 juillet 2026":"Official model notes, July 9, 2026","Nous utilisons Google Analytics et Vercel Analytics pour mesurer la fréquentation et les actions sur le site, avec votre accord. Votre choix vaut 180 jours. Vous pouvez le retirer à tout moment.":"We use Google Analytics and Vercel Analytics to measure visits and actions on the site, with your consent. Your choice lasts 180 days. You can withdraw it at any time.","Nutzungsmessung":"Usage measurement","NVIDIA rend disponibles le Toolkit et le SDK CUDA 1.0 pour programmer ses GPU au-delà du rendu graphique.":"NVIDIA makes the CUDA 1.0 Toolkit and SDK available for programming its GPUs beyond graphics rendering.","o1 introduit une série dédiée au raisonnement":"o1 introduces a series dedicated to reasoning","Offline Memory conserve et recherche les souvenirs des agents dans une base SQLite. Il associe une recherche lexicale FTS5/BM25, un classement sémantique optionnel et une attribution par entité et session.":"Offline Memory stores and searches agent memories in a SQLite database. It combines FTS5/BM25 lexical search, optional semantic ranking and attribution by entity and session.","Okodukai est une tirelire virtuelle pour apprendre à gérer son argent. L'enfant gagne des pièces grâce aux quêtes proposées par ses parents, puis choisit de les dépenser dans la boutique familiale, de les garder dans son Coffre magique ou de découvrir les placements dans une simulation.":"Okodukai is a virtual piggy bank for learning to manage money. Children earn coins through quests proposed by their parents, then choose to spend them in the family shop, keep them in their magic chest or explore investments in a simulation.","On remonte le temps...":"Travelling back in time...","On se prend":"Shall we pick","OpenAI annonce GPT-4o, un modèle réunissant traitement du texte, de l'image et de l'audio.":"OpenAI announces GPT-4o, a model combining text, image and audio processing.","OpenAI annonce son laboratoire de recherche":"OpenAI announces its research lab","OpenAI annonce une réfutation de la conjecture des distances unité d'Erdős":"OpenAI announces a refutation of Erdős's unit-distance conjecture","OpenAI associe préentraînement génératif et ajustement supervisé pour améliorer plusieurs tâches de compréhension du langage.":"OpenAI combines generative pretraining and supervised fine-tuning to improve several language understanding tasks.","OpenAI commence le déploiement de GPT-5.6 Sol pour les tâches complexes de code, recherche, sciences et utilisation de l'ordinateur.":"OpenAI begins rolling out GPT-5.6 Sol for complex coding, research, science and computer-use tasks.","OpenAI déploie GPT-6 avec Intelligent UI, pour composer des réponses associant texte, éléments visuels et interactions.":"OpenAI rolls out GPT-6 with Intelligent UI to compose answers combining text, visuals and interactions.","OpenAI étend Agents API à l'interaction avec des logiciels et aux tâches multi-agents sur son infrastructure.":"OpenAI expands Agents API to software interaction and multiagent tasks on its infrastructure.","OpenAI lance Dots, ses agents personnels persistants":"OpenAI launches Dots, its persistent personal agents","OpenAI lance GPT-5 comme système unifié capable de choisir entre réponse rapide et raisonnement approfondi.":"OpenAI launches GPT-5 as a unified system that can choose between fast answers and in-depth reasoning.","OpenAI lance GPT-5.5 pour les tâches de code, de recherche et de travail sur ordinateur.":"OpenAI launches GPT-5.5 for coding, research and computer work.","OpenAI lance Sol et Luna pour proposer différents compromis entre capacités, rapidité et coût dans la famille GPT-6.":"OpenAI launches Sol and Luna to offer different balances of capability, speed and cost in the GPT-6 family.","OpenAI lance Sora 2, avec génération audio synchronisée et une nouvelle application.":"OpenAI launches Sora 2, with synchronized audio generation and a new app.","OpenAI lance une bêta privée de son API permettant d'intégrer des modèles de la famille GPT-3 dans des applications.":"OpenAI launches a private beta of its API for integrating GPT-3 family models into applications.","OpenAI ouvre DALL·E 3 aux abonnés ChatGPT Plus et Enterprise pour générer des images par conversation.":"OpenAI opens DALL·E 3 to ChatGPT Plus and Enterprise subscribers for conversational image generation.","OpenAI ouvre o1-preview et o1-mini, conçus pour consacrer davantage de calcul au raisonnement avant de répondre.":"OpenAI opens o1-preview and o1-mini, designed to spend more compute on reasoning before answering.","OpenAI ouvre Sora Turbo comme produit de génération vidéo aux abonnés ChatGPT Plus et Pro dans les zones prises en charge.":"OpenAI opens Sora Turbo as a video generation product for ChatGPT Plus and Pro subscribers in supported regions.","OpenAI ouvre un dépôt de manuscrits produits avec un modèle interne, avec des formalisations Lean et des informations sur la méthode de recherche.":"OpenAI opens a repository of manuscripts produced with an internal model, with Lean formalizations and information on the research method.","OpenAI ouvre une préversion de Codex, agent cloud capable de modifier du code et d'exécuter des tests.":"OpenAI opens a preview of Codex, a cloud agent capable of modifying code and running tests.","OpenAI présente Decisions API":"OpenAI introduces Decisions API","OpenAI présente des modèles ajustés avec des démonstrations et des préférences humaines, devenus les modèles par défaut de son API.":"OpenAI introduces models tuned with human demonstrations and preferences, which become the default models in its API.","OpenAI présente GPT-2 et publie d'abord une version plus petite, avant de diffuser les versions suivantes.":"OpenAI introduces GPT-2 and initially releases a smaller version before releasing subsequent versions.","OpenAI présente GPT-4 et ouvre son accès dans ChatGPT Plus ainsi qu'en API sur liste d'attente.":"OpenAI introduces GPT-4 and opens access in ChatGPT Plus and through an API waitlist.","OpenAI présente les recherches derrière Sora, capable de générer des vidéos à partir de texte.":"OpenAI presents the research behind Sora, capable of generating videos from text.","OpenAI propose une preuve pour le problème du millénaire de Navier-Stokes":"OpenAI proposes a proof for the Navier-Stokes Millennium Prize Problem","OpenAI publie dix résultats de mathématiques et d'informatique théorique":"OpenAI publishes ten results in mathematics and theoretical computer science","OpenAI publie GPT-5.4 dans ChatGPT, Codex et l'API, avec des capacités natives d'utilisation de l'ordinateur.":"OpenAI releases GPT-5.4 in ChatGPT, Codex and the API, with native computer-use capabilities.","OpenAI publie GPT-6 Astra":"OpenAI releases GPT-6 Astra","OpenAI publie GPT-6 Astra, avec des capacités renforcées de travail professionnel et d'utilisation de l'ordinateur.":"OpenAI releases GPT-6 Astra, with stronger professional work and computer-use capabilities.","OpenAI publie gpt-oss-120b et gpt-oss-20b, deux modèles de raisonnement à poids ouverts sous Apache 2.0.":"OpenAI releases gpt-oss-120b and gpt-oss-20b, two open-weight reasoning models under Apache 2.0.","OpenAI publie les modèles et le code de Whisper pour la reconnaissance vocale et la traduction de la parole vers l'anglais.":"OpenAI releases Whisper's models and code for speech recognition and speech translation into English.","OpenAI publie les poids de gpt-oss":"OpenAI releases the gpt-oss weights","OpenAI publie un corpus de résultats sur des problèmes mathématiques ouverts":"OpenAI publishes a corpus of results on open mathematical problems","OpenAI publie un résultat de géométrie discrète obtenu par un modèle interne, présenté comme une réfutation de la conjecture des distances unité d'Erdős.":"OpenAI publishes a discrete geometry result obtained by an internal model, presented as a refutation of Erdős's unit-distance conjecture.","OpenAI publie une preuve analytique et une formalisation Lean d'une singularité en temps fini pour un fluide 3D soumis à une force extérieure lisse.":"OpenAI publishes an analytical proof and Lean formalization of a finite-time singularity for a 3D fluid subject to a smooth external force.","OpenAI se présente publiquement comme une organisation de recherche en IA à but non lucratif.":"OpenAI publicly introduces itself as a nonprofit AI research organization.","Operator ouvre une préversion de navigation autonome":"Operator opens an autonomous browsing preview","Operator utilise son propre navigateur pour cliquer, saisir du texte et réaliser des tâches sur des sites web.":"Operator uses its own browser to click, enter text and perform tasks on websites.","Option inconnue. Consulter --help.":"Unknown option. See --help.","Option répétée. Consulter --help.":"Repeated option. See --help.","organisations représentées":"organizations represented","Origine du site invalide.":"Invalid site origin.","Origine vidéo non autorisée":"Video origin not allowed","OUVERT":"OPEN","Ouvert aux curieux.":"Open to the curious.","Ouverture / sortie":"Access / release","Ouverture de DALL·E 3 dans ChatGPT":"DALL·E 3 opens in ChatGPT","Ouverture de la boîte":"Opening the box","Ouverture de la préversion de Codex":"Codex preview opens","Ouverture publique de Deep Review":"Deep Review opens to the public","Ouvrir":"Open","Ouvrir la boîte":"Open the box","Ouvrir la fiche complète":"Open the full entry","PAL":"PAL","PAL /":"PAL /","PaLM pousse l'échelle à 540 milliards de paramètres":"PaLM scales to 540 billion parameters","Par":"By","Parcourir les événements chronologiques":"Browse chronological events","Partager cette année":"Share this year","Partager cette date":"Share this date","Passionné par les jeux vidéo et les nouvelles technologies, Lou explore comment appliquer les mécaniques du jeu vidéo à l'éducation et à la pédagogie. Son ambition : transformer les programmes scolaires et les apprentissages du quotidien en expériences ludiques et interactives, pour apprendre autrement, en mêlant l'utile à l'agréable. À travers des projets comme Héros de la classe, Okodukai ou Not My Fault, il expérimente de nouvelles façons d'apprendre, de collaborer et de développer ses compétences en jouant.":"Passionate about video games and new technologies, Lou explores how game mechanics can support education and teaching. His ambition is to turn school curricula and everyday learning into playful, interactive experiences, combining usefulness and enjoyment to learn differently. Through projects such as Héros de la classe, Okodukai and Not My Fault, he experiments with new ways to learn, collaborate and build skills through play.","PÉRIODE AFFICHÉE":"DISPLAYED PERIOD","Période inversée":"Reversed period","Période ou catégorie incompatible avec le type de récompense.":"Period or category incompatible with the award type.","Périodes de l'histoire de l'IA":"Periods in AI history","Pilotez le retour de Super Heavy.":"Pilot Super Heavy's return.","PinchChat permet de suivre les réponses et les actions de son agent OpenClaw en temps réel. Il réunit les conversations, tâches de fond et sessions d'agents dans une interface personnalisable.":"PinchChat lets you follow your OpenClaw agent's answers and actions in real time. It brings conversations, background tasks and agent sessions together in a customizable interface.","Planifier les prochaines tâches":"Plan upcoming tasks","Plateau de Sandbox":"Sandbox studio","Plateforme":"Platform","Plateformes":"Platforms","PLAY":"PLAY","Plus anciens d'abord":"Oldest first","Plus récents d'abord":"Newest first","Podcasts IA":"AI podcasts","Point Zero représente une explosion nucléaire en trois dimensions, à l'endroit choisi sur la carte. La simulation affiche les zones d'impact et donne un repère visuel sur l'échelle du phénomène.":"Point Zero depicts a nuclear explosion in three dimensions at the chosen map location. The simulation shows impact zones and provides a visual reference for the scale of the phenomenon.","Politiquizz rend les données politiques françaises plus faciles à explorer. Les quiz proposent de répondre à des questions de fond, puis de comparer ses réponses aux votes des groupes à l'Assemblée nationale ou aux propositions des programmes.":"Politiquizz makes French political data easier to explore. Quizzes ask substantive questions, then compare answers with National Assembly groups' votes or manifesto proposals.","Précédent":"Previous","Preferencias de medición":"Measurement preferences","Premier épisode (pilote) de Sandbox, le podcast qui teste, débat et remet les chiffres en perspective. Lou, Nicolas et Loïc passent en revue l'actualité IA et tech de la semaine, présentent un outil fait maison et racontent leurs expérimentations en cours.":"First pilot episode of Sandbox, the podcast that tests, debates and puts numbers into perspective. Lou, Nicolas and Loïc review the week's AI and tech news, present a home-built tool and discuss their ongoing experiments.","Premiers pas avec Hivekeep":"Getting started with Hivekeep","Préprint original sur AlphaZero":"Original AlphaZero preprint","Présentation d'AlphaEvolve":"Introducing AlphaEvolve","Présentation de AI co-scientist":"Introducing AI co-scientist","Présentation de Akasha":"Introducing Akasha","Présentation de Akasha Model":"Introducing Akasha Model","Présentation de Akasha OS":"Introducing Akasha OS","Présentation de BallCam.tv":"Introducing BallCam.tv","Présentation de Bot Chronicles":"Introducing Bot Chronicles","Présentation de Claude":"Introducing Claude","Présentation de Claude Haiku 5.5":"Introducing Claude Haiku 5.5","Présentation de Claude Opus 5.5":"Introducing Claude Opus 5.5","Présentation de Claude Sonnet 5.5":"Introducing Claude Sonnet 5.5","Présentation de Gemini":"Introducing Gemini","Présentation de Gemini 2.0":"Introducing Gemini 2.0","Présentation de Gemini Robotics":"Introducing Gemini Robotics","Présentation de GPT-6.1 Sol":"Introducing GPT-6.1 Sol","Présentation de Héros de la classe":"Introducing Héros de la classe","Présentation de Hivekeep":"Introducing Hivekeep","Présentation de MindOdyssey":"Introducing MindOdyssey","Présentation de Not My Fault":"Introducing Not My Fault","Présentation de Okodukai":"Introducing Okodukai","Présentation de PinchChat":"Introducing PinchChat","Présentation de Point Zero":"Introducing Point Zero","Présentation de Politiquizz":"Introducing Politiquizz","Présentation de Raycast Uptime Kuma":"Introducing Raycast Uptime Kuma","Présentation de Rbitnet":"Introducing Rbitnet","Présentation de RT-2":"Introducing RT-2","Présentation de Sillage":"Introducing Sillage","Présentation de Song Maker":"Introducing Song Maker","Présentation de Streamdeck Uptime Kuma":"Introducing Streamdeck Uptime Kuma","Présentation des travaux Aletheia":"Presentation of Aletheia's work","Présentation et publication d'AlphaGeometry":"AlphaGeometry presentation and publication","Présentation initiale de Devin":"Initial Devin presentation","Présentation par Nicolas sur LinkedIn":"Presentation by Nicolas on LinkedIn","Preuve proposée et précisions d'OpenAI":"Proposed proof and clarifications from OpenAI","Preuves de cette édition":"This edition's evidence","Privacidad y plazos de conservación":"Privacy and retention periods","Privacy and retention periods":"Privacy and retention periods","product":"product","Produit":"Product","Projet absent.":"Missing project.","Projet en double":"Duplicate project","PROJETS EN RAYON":"PROJECTS ON THE SHELVES","Projets open source, skills et modèles IA, avec éditions datées et méthodes publiques.":"Open-source projects, skills and AI models, with dated editions and public methods.","Protocole et premiers résultats AA-Robotics":"AA-Robotics protocol and initial results","Publication AlphaMissense dans Science":"AlphaMissense publication in Science","Publication d'o1-preview":"o1-preview release","Publication datée du partenaire Hugging Face":"Dated publication from partner Hugging Face","Publication de Claude 3.7 et Claude Code":"Claude 3.7 and Claude Code release","Publication de Claude 4":"Claude 4 release","Publication de DeepSeek-R1":"DeepSeek-R1 release","Publication de DeepSeek-V3":"DeepSeek-V3 release","Publication de DeepSeek-V4 Preview":"DeepSeek-V4 Preview release","Publication de FLUX.1":"FLUX.1 release","Publication de FunSearch":"FunSearch publication","Publication de Gemini 2.5":"Gemini 2.5 release","Publication de Gemini 3.1 Pro":"Gemini 3.1 Pro release","Publication de GLM-5.3":"GLM-5.3 release","Publication de GPT-5":"GPT-5 release","Publication de GPT-5.4":"GPT-5.4 release","Publication de GPT-5.5":"GPT-5.5 release","Publication de GraphCast":"GraphCast publication","Publication de Llama 3":"Llama 3 release","Publication de Llama 3.1":"Llama 3.1 release","Publication de Llama 4":"Llama 4 release","Publication de Mistral 3":"Mistral 3 release","Publication de Mistral 7B":"Mistral 7B release","Publication de Mixtral":"Mixtral release","Publication de Qwen3":"Qwen3 release","Publication de recherche AlphaProteo":"AlphaProteo research publication","Publication de Sora 2":"Sora 2 release","Publication des dix résultats":"Publication of the ten results","Publication des modèles gpt-oss":"gpt-oss model release","Publication du Model Context Protocol":"Model Context Protocol release","Publication et présentation d'AlphaTensor":"AlphaTensor publication and presentation","Publication et rapport technique HiRO-ACEv2":"HiRO-ACEv2 release and technical report","Publication finale des poids de GPT-2":"Final release of GPT-2 weights","Publication originale, Nature":"Original publication, Nature","Publication scientifique":"Scientific publication","Publication scientifique AlphaDev":"AlphaDev scientific publication","Publication sur la conjecture d'Erdős":"Publication on Erdős's conjecture","Published methodology unavailable:":"Published methodology unavailable:","PyTorch devient accessible aux chercheurs":"PyTorch becomes accessible to researchers","PyTorch ouvre un environnement Python de calcul sur tenseurs et d'entraînement de réseaux avec des graphes dynamiques.":"PyTorch opens a Python environment for tensor computation and network training with dynamic graphs.","Qwen publie huit modèles aux poids ouverts sous Apache 2.0, avec des modes de réponse rapide et de raisonnement.":"Qwen releases eight open-weight models under Apache 2.0, with fast-answer and reasoning modes.","Qwen3 publie une famille multilingue à raisonnement hybride":"Qwen3 releases a multilingual family with hybrid reasoning","Qwen3.8-Max ouvre son accès API":"Qwen3.8-Max opens API access","Raisonnement & agents":"Reasoning & agents","Rappel officiel de la date de lancement":"Official reminder of the launch date","Rapport et accès expérimentaux Earth AI":"Earth AI report and experimental access","Rapport initial de Sora":"Initial Sora report","Rayon précédent":"Previous shelf","Rayon suivant":"Next shelf","Rbitnet fournit un moteur d'inférence local écrit en Rust et un serveur HTTP compatible avec l'API OpenAI. Il charge des modèles GGUF selon les architectures prises en charge dans sa documentation.":"Rbitnet provides a local inference engine written in Rust and an HTTP server compatible with the OpenAI API. It loads GGUF models according to the architectures supported in its documentation.","Récapitulatif officiel du DevDay 2026":"Official DevDay 2026 recap","Rechazar":"Reject","Recherche et prépublication sur les enzymes ART":"Research and preprint on ART enzymes","Rechercher une annonce, un modèle ou une organisation":"Search for an announcement, model or organization","Rechercher une cassette":"Search for a tape","Reconnaître, apprendre, jouer":"Recognize, learn, play","Réessayer":"Try again","Refuser":"Reject","Reject":"Reject","Rejouer l'histoire":"Replay history","Relire les modifications Git":"Review Git changes","REMBOBINER":"REWIND","REMONTEZ LE TEMPS":"TRAVEL BACK IN TIME","Replays Sandbox":"Sandbox replays","Réponse HTML attendue":"HTML response expected","Reposer":"Put back","Reposer la cassette":"Put the tape back","research":"research","ResNet facilite les réseaux très profonds":"ResNet enables very deep networks","Responsable, services de mesure, durées de conservation et retrait du consentement sur Sandbox.":"Controller, measurement services, retention periods and consent withdrawal on Sandbox.","Ressource absente.":"Missing resource.","Retirar mi consentimiento":"Withdraw my consent","Retirer mon accord":"Withdraw my consent","Retour au plateau":"Back to the studio","Retrait : --id doit être un UUID valide, utilisé sans --week.":"Withdrawal: --id must be a valid UUID used without --week.","Retrait : --id ou --week valide, --reason de 10 à 1000 caractères et --actor de 1 à 100 caractères requis.":"Withdrawal: valid --id or --week, --reason of 10 to 1000 characters and --actor of 1 to 100 characters required.","RÉTROSPECTIVE":"RETROSPECTIVE","Retrouver les notes":"Find notes","Retrouvez les animateurs de Sandbox et leurs liens publics.":"Find the Sandbox hosts and their public links.","Revenir au premier événement":"Return to the first event","Revue au":"Reviewed on","RFdiffusion conçoit de nouvelles protéines":"RFdiffusion designs new proteins","RISING STAR":"RISING STAR","Rookie of the Month":"Rookie of the Month","ROOKIE OF THE MONTH":"ROOKIE OF THE MONTH","Router les requêtes entre plusieurs modèles.":"Route requests between multiple models.","RT-2 relie langage, vision et action robotique":"RT-2 connects language, vision and robotic action","RT-2 transpose des connaissances apprises sur le Web en commandes de contrôle robotique.":"RT-2 transfers knowledge learned from the Web into robotic control commands.","Rust Commander permet d'administrer un serveur du jeu Rust depuis Slack ou Discord. Le bot relaie les conversations et les logs et transmet des commandes à la console.":"Rust Commander lets you manage a Rust game server from Slack or Discord. The bot relays conversations and logs and sends commands to the console.","Sandbox #1 (pilote) Mistral Large 4, Claude Haiku 5.5, Gemini at Work, Sillage et les harnais IA":"Sandbox #1 (pilot) Mistral Large 4, Claude Haiku 5.5, Gemini at Work, Sillage and AI harnesses","Sandbox Annual Awards":"Sandbox Annual Awards","Sandbox award edition unavailable:":"Sandbox award edition unavailable:","Sandbox award metadata unavailable:":"Sandbox award metadata unavailable:","Sandbox award unavailable:":"Sandbox award unavailable:","Sandbox Awards":"Sandbox Awards","Sandbox Awards history unavailable:":"Sandbox Awards history unavailable:","SANDBOX BADGES & AWARDS":"SANDBOX BADGES & AWARDS","Sandbox entity awards unavailable:":"Sandbox entity awards unavailable:","Sandbox entity metadata unavailable:":"Sandbox entity metadata unavailable:","Sandbox entity unavailable:":"Sandbox entity unavailable:","SANDBOX HOME VIDEO":"SANDBOX HOME VIDEO","SANDBOX PRÉSENTE":"SANDBOX PRESENTS","Sandbox Rising Star":"Sandbox Rising Star","SANDBOX RISING STAR /":"SANDBOX RISING STAR /","Sandbox Rising Star of the Week":"Sandbox Rising Star of the Week","Sandbox Top 100":"Sandbox Top 100","Sandbox Vidéo Club":"Sandbox Video Club","Sandbox Vidéo Club (essai local)":"Sandbox Video Club (local trial)","SANDBOX WRAPPED":"SANDBOX WRAPPED","sandbox.podcast.pro@gmail.com":"sandbox.podcast.pro@gmail.com","science":"science","Sciences":"Science","SDXL 1.0 est publié":"SDXL 1.0 is released","Se déplacer dans le vidéoclub":"Move around the video club","SÉLECTION ÉDITORIALE DE L'ANNÉE":"EDITORIAL SELECTION OF THE YEAR","Sélection éditoriale non exhaustive de l'histoire mondiale de l'IA.":"A non-exhaustive editorial selection from the world history of AI.","Sélection non exhaustive, vérifiée auprès des laboratoires, des chercheurs et des institutions. La courbe cumule les événements retenus.":"Non-exhaustive selection, verified with labs, researchers and institutions. The curve accumulates the selected events.","Sélection non exhaustive.":"Non-exhaustive selection.","Semaine ISO inexistante.":"ISO week does not exist.","Semaine ISO invalide. Format attendu : AAAA-Wnn.":"Invalid ISO week. Expected format: AAAA-Wnn.","Sequence to Sequence Learning with Neural Networks":"Sequence to Sequence Learning with Neural Networks","Sillage en 54 secondes":"Sillage in 54 seconds","Sillage réunit Claude Code, Codex et OpenCode dans un espace de travail à héberger sur sa machine ou son serveur. Les agents utilisent leurs outils natifs et vos comptes. Le navigateur donne accès aux conversations, aux projets et aux demandes de permission, depuis un ordinateur comme depuis un téléphone.":"Sillage brings Claude Code, Codex and OpenCode into a workspace hosted on your machine or server. Agents use their native tools and your accounts. The browser provides access to conversations, projects and permission requests from a computer or phone.","Siri arrive dans l'iPhone":"Siri arrives on the iPhone","Site web":"Website","Sitemap urlset XML attendu.":"Sitemap XML urlset expected.","Skill of the Week":"Skill of the Week","SKILL OF THE WEEK":"SKILL OF THE WEEK","Société & règles":"Society & rules","Son usage principal dans Akasha OS est le contrôle d'un appel d'outil avant son exécution : autoriser, s'abstenir ou bloquer. Le dépôt contient des démonstrations textuelles et visuelles. Les extraits de jeu présentés sont des exemples choisis, pas une mesure de performances typiques.":"Its main use in Akasha OS is checking a tool call before execution: allow, abstain or block. The repository contains text and visual demonstrations. The game clips shown are selected examples, not a measure of typical performance.","Song Maker est un atelier musical local. Un style et des paroles servent à générer une chanson avec YuE2. Les prises peuvent ensuite être comparées, éditées et séparées en pistes avec HTDemucs.":"Song Maker is a local music workshop. A style and lyrics generate a song with YuE2. Takes can then be compared, edited and separated into stems with HTDemucs.","Sora 2 associe génération vidéo et audio":"Sora 2 combines video and audio generation","Sora montre la vidéo générée par IA":"Sora showcases AI-generated video","Sora sort de la préversion de recherche":"Sora leaves the research preview","Source HTTPS requise":"HTTPS source required","Source primaire :":"Primary source:","Sources et méthode":"Sources and method","sources primaires":"primary sources","Space et Pages proposent un espace partagé pour travailler avec ChatGPT, les Dots et des collègues.":"Space and Pages offer a shared workspace with ChatGPT, Dots and colleagues.","Spotify":"Spotify","Stability AI publie les poids et le code de SDXL 1.0, son modèle de génération d'images.":"Stability AI publishes the weights and code for SDXL 1.0, its image generation model.","Stable Diffusion devient téléchargeable":"Stable Diffusion becomes downloadable","Stable Diffusion est diffusé publiquement avec ses poids et son code, permettant la génération d'images sur du matériel local compatible.":"Stable Diffusion is publicly released with its weights and code, enabling image generation on compatible local hardware.","Stanley remporte le DARPA Grand Challenge":"Stanley wins the DARPA Grand Challenge","Suivant":"Next","Suivre une mission":"Follow a mission","Sutskever, Vinyals et Le présentent un système de réseaux LSTM qui encode une phrase puis produit sa traduction.":"Sutskever, Vinyals and Le present an LSTM network system that encodes a sentence and then produces its translation.","T5 traite traduction, résumé et questions-réponses dans un même format où l'entrée et la sortie sont du texte.":"T5 handles translation, summarization and question answering in a common format where input and output are text.","T5 transforme les tâches de langage en texte vers texte":"T5 turns language tasks into text-to-text tasks","Team Bots partage les agents Grok au sein d'une équipe":"Team Bots shares Grok agents within a team","Team Bots permet de partager un agent Grok doté de fichiers, de compétences, de connexions aux applications et d'une mémoire commune.":"Team Bots lets teams share a Grok agent with files, skills, app connections and shared memory.","TeamSpeak MCP relie un assistant IA à un serveur TeamSpeak. Il peut consulter les utilisateurs et les canaux, envoyer des messages et exécuter les opérations d'administration exposées par le serveur.":"TeamSpeak MCP connects an AI assistant to a TeamSpeak server. It can view users and channels, send messages and execute administration operations exposed by the server.","téléchargé.":"downloaded.","Télécharger":"Download","Télécharger le Wrapped":"Download the Wrapped","TensorFlow devient public":"TensorFlow becomes public","Terence Tao explique un contre-exemple à la conjecture jacobienne obtenu avec Fable":"Terence Tao explains a counterexample to the Jacobian conjecture obtained with Fable","Terence Tao expose et vérifie un contre-exemple polynomial en dimension trois à la conjecture jacobienne, obtenu avec l'aide de Fable.":"Terence Tao presents and verifies a three-dimensional polynomial counterexample to the Jacobian conjecture, obtained with Fable's help.","Texte intégral de l'article DeepFace":"Full text of the DeepFace paper","Texte, image, son et vidéo":"Text, image, sound and video","The X Subscription rassemble Grok, Grok Bot, Cursor et X Premium dans une offre commune.":"The X Subscription brings Grok, Grok Bot, Cursor and X Premium together in one offering.","Thing":"Thing","Three Observers compare le ciel vu depuis trois endroits de la Terre au même instant. La simulation affiche côte à côte un globe et un modèle de Terre plate pour observer les conséquences de leur géométrie sur le mouvement apparent des étoiles.":"Three Observers compares the sky seen from three locations on Earth at the same instant. The simulation shows a globe alongside a flat-Earth model to observe how their geometry affects the apparent motion of stars.","Time Machine":"Time Machine","TITRES":"TITLES","TITRES / VHS":"TITLES / VHS","tools":"tools","Tout afficher":"Show all","Toutes les années":"All years","Toutes les annonces":"All announcements","Toutes les annonces de":"All announcements from","Toutes les cassettes":"All tapes","Training Compute-Optimal Large Language Models":"Training Compute-Optimal Large Language Models","Travailler sur le code avec les agents Akasha.":"Work on code with Akasha agents.","TU AS CHOISI":"YOU CHOSE","Type de récompense inconnu. Consulter --help.":"Unknown award type. See --help.","Types d'événements":"Event types","Un abonnement X réunit Grok, Grok Bot et Cursor":"An X subscription bundles Grok, Grok Bot and Cursor","Un agent combine réseau profond et apprentissage par renforcement pour jouer à 49 jeux à partir des pixels et du score.":"An agent combines a deep network and reinforcement learning to play 49 games using pixels and scores.","Un aperçu de l'atelier musical":"A glimpse of the music workshop","Un assistant IA local, avec mémoire.":"A local AI assistant with memory.","Un candidat médicament conçu avec l'IA publie ses résultats de phase 2a":"An AI-designed drug candidate publishes its phase 2a results","Un chat créé avec Cursor.":"A chat built with Cursor.","Un compagnon vocal pour dialoguer avec Akasha.":"A voice companion for talking with Akasha.","Un contrôle autorisé":"An allowed control","Un contrôle bloqué":"A blocked control","Un espace de travail pour vos agents de code.":"A workspace for your coding agents.","Un même algorithme apprend séparément trois jeux par autojeu et bat des programmes de référence dans les évaluations publiées.":"The same algorithm learns three games separately through self-play and beats reference programs in published evaluations.","Un modèle repère une molécule dont l'activité antibiotique est ensuite testée en laboratoire et chez la souris.":"A model identifies a molecule whose antibiotic activity is then tested in the laboratory and in mice.","Un modèle, une annonce...":"A model, an announcement...","Un moteur d'inférence local écrit en Rust.":"A local inference engine written in Rust.","UN PROJET.":"ONE PROJECT.","Un relais de conversation entre agents.":"A conversation relay between agents.","Un système d'apprentissage par renforcement pilote les bobines magnétiques du tokamak TCV pour maintenir différentes formes de plasma.":"A reinforcement learning system controls the magnetic coils of the TCV tokamak to maintain different plasma shapes.","Un tableau pour suivre le travail":"A board to track work","Un titre en tête ?":"Have a title in mind?","une cassette ?":"a tape?","Une collaboration internationale publie BLOOM, un modèle de 176 milliards de paramètres couvrant 46 langues naturelles et 13 langages de programmation.":"An international collaboration releases BLOOM, a 176-billion-parameter model covering 46 natural languages and 13 programming languages.","Une création de":"Created by","Une crise. Des indices. Votre équipe.":"A crisis. Clues. Your team.","Une démonstration sur Tetris":"A Tetris demonstration","Une équipe d'agents IA, hébergée chez vous.":"A team of AI agents hosted by you.","Une IA contrôle le plasma d'un tokamak":"An AI controls tokamak plasma","Une interface de conversation pour OpenClaw.":"A chat interface for OpenClaw.","Une mémoire locale pour les agents OpenClaw.":"Local memory for OpenClaw agents.","Une méthode de diffusion génère des structures de protéines, notamment des assemblages et des protéines capables de se lier à une cible.":"A diffusion method generates protein structures, including assemblies and proteins capable of binding to a target.","Une méthode pour entraîner des réseaux profonds":"A method for training deep networks","Une notification peut viser un appareil ou tous les appareils, dans la langue choisie. Ce projet ancien reste disponible en code source. Son README signale un fonctionnement sur macOS et des problèmes de dépendances sur Windows et Linux.":"A notification can target one device or all devices, in the selected language. This older project remains available as source code. Its README reports operation on macOS and dependency issues on Windows and Linux.","Une partie coopérative réunit deux à cinq joueurs dans le navigateur. Le mode PvP oppose deux équipes sur une même mission. Le lien de la salle se partage aux participants, sans compte obligatoire ni installation. Un appel vocal habituel peut accompagner la partie.":"A cooperative game brings together two to five players in the browser. PvP mode pits two teams against each other on the same mission. Share the room link with participants, with no mandatory account or installation. A regular voice call can accompany the game.","UNE SOIRÉE.":"ONE EVENING.","Une version avancée de Gemini Deep Think résout cinq des six problèmes de l'IMO 2025 et obtient 35 points sur 42.":"An advanced version of Gemini Deep Think solves five of the six IMO 2025 problems and scores 35 points out of 42.","URL dupliquée dans le sitemap.":"Duplicate URL in sitemap.","URL HTTP sans identifiants ni fragment requise":"HTTP URL without credentials or fragment required","URL non indexable (robots)":"URL not indexable (robots)","Usamos Google Analytics y Vercel Analytics para medir el uso del sitio. Estos servicios solo se activan con tu consentimiento. Puedes cambiar tu elección en cualquier momento.":"We use Google Analytics and Vercel Analytics to measure site usage. These services are enabled only with your consent. You can change your choice at any time.","Veo 3 génère la vidéo avec son":"Veo 3 generates video with sound","Vérification publique indisponible.":"Public verification unavailable.","Vérifié le":"Verified on","VHS":"VHS","VIDÉO CLUB":"VIDEO CLUB","Vidéo requise":"Video required","Vidéoclub. Flèches gauche et droite pour changer de rayon, haut pour avancer, bas pour reculer.":"Video club. Left and right arrows change shelves, up moves forward, down moves backward.","Voir la liste":"View the list","Voir les rayons":"View the shelves","Voir les souvenirs et l'activité de son agent.":"View your agent's memories and activity.","Vos DNS suivent vos conteneurs Docker.":"Your DNS follows your Docker containers.","Vos moniteurs Uptime Kuma dans Raycast.":"Your Uptime Kuma monitors in Raycast.","Vos replays Rocket League, sous tous les angles.":"Your Rocket League replays, from every angle.","Votre choix n’a pas pu être enregistré. Autorisez le stockage local puis réessayez.":"Your choice could not be saved. Allow local storage and try again.","VOTRE CHOIX.":"YOUR CHOICE.","Votre navigateur ne peut pas lire cette vidéo.":"Your browser cannot play this video.","Vue d'ensemble":"Overview","Wan2.1 publie code et poids pour la vidéo":"Wan2.1 releases code and weights for video","Watson bat Ken Jennings et Brad Rutter dans un concours télévisé fondé sur des questions en langage naturel.":"Watson beats Ken Jennings and Brad Rutter in a television quiz based on natural-language questions.","Watson gagne Jeopardy!":"Watson wins Jeopardy!","WaveNet génère directement le signal audio":"WaveNet generates the audio signal directly","WaveNet produit la parole en modélisant les échantillons du signal sonore, plutôt qu'en assemblant des fragments enregistrés.":"WaveNet produces speech by modeling sound samples rather than assembling recorded fragments.","We use Google Analytics and Vercel Analytics to measure site traffic. These services are enabled only with your consent. You can change your choice at any time.":"We use Google Analytics and Vercel Analytics to measure site traffic. These services are enabled only with your consent. You can change your choice at any time.","Whisper ouvre la transcription multilingue":"Whisper opens multilingual transcription","Wir verwenden Google Analytics und Vercel Analytics, um die Nutzung der Website zu messen. Diese Dienste werden erst nach Ihrer Zustimmung aktiviert. Sie können Ihre Auswahl jederzeit ändern.":"We use Google Analytics and Vercel Analytics to measure site usage. These services are enabled only with your consent. You can change your choice at any time.","Withdraw consent":"Withdraw consent","word2vec représente les mots par des vecteurs":"word2vec represents words as vectors","WRAPPED":"WRAPPED","x":"x","xAI annonce Grok 4.7, une nouvelle version de sa famille de modèles destinée aux usages de Grok et de ses agents.":"xAI announces Grok 4.7, a new version of its model family for Grok and its agents.","Your choice could not be saved. Allow local storage and try again.":"Your choice could not be saved. Allow local storage and try again.","YouTube":"YouTube","Z.ai publie GLM-5.3, issu du même modèle de base que GLM-5.2 avec un entraînement complémentaire centré sur les tâches longues.":"Z.ai releases GLM-5.3, based on the same base model as GLM-5.2 with additional training focused on long tasks.","Consultez les classements publiés ou les épisodes du Talk Show.":"Browse the published rankings or Talk Show episodes.","Voir les épisodes":"View episodes","{topic} : les épisodes du Talk Show":"{topic}: Talk Show episodes","— Talk Show et classements tech":"— Video Talk Show and tech rankings","Dans le Talk Show":"In the Talk Show","Sandbox, le Talk Show sur l'IA et la tech. Retrouvez les épisodes, leurs sources et les classements commentés.":"Sandbox is a video Talk Show about AI and tech. Explore the episodes, their sources and the charts with commentary.","Le Talk Show qui teste, débat et remet les chiffres en perspective.":"The Talk Show that tests, debates and puts figures into perspective.","Les épisodes":"Episodes","Plateformes du Talk Show":"Talk Show platforms","Talk Show indépendant · IA · développement · open source":"Independent Talk Show · AI · development · open source","Rechercher dans le Talk Show et les classements":"Search Talk Show and rankings","Tous les épisodes.":"All podcasts.","Un Talk Show pour y voir plus clair.":"A Talk Show to make things clearer.","Talk Show et classements IA":"Talk Show and AI rankings","Sandbox · Talk Show":"Sandbox · Talk Show","Search the talk show and AI rankings":"Search Talk Show and AI rankings","DANS LE TALK SHOW":"IN THE TALK SHOW","L'équipe du Talk Show Sandbox est responsable des traitements décrits ici. Pour une question ou l'exercice de vos droits, écrivez à":"The Sandbox Talk Show team is responsible for the processing described here. For questions or to exercise your rights, write to","Les avis de l'équipe sont signés et séparés du calcul. Une recommandation dans le Talk Show ne modifie pas un rang.":"The team's opinions are signed and separate from the calculation. A Talk Show recommendation does not change a rank.","Le Talk Show":"The Talk Show","Le Talk Show Sandbox : replays et historique":"The Sandbox Talk Show: replays and archive","Le Talk Show de Lou, Loïc et Nico":"The Talk Show with Lou, Loïc and Nico","Le Talk Show IA & tech":"The AI & tech talk show","Le Talk Show IA & tech de Lou, Loïc et Nico.":"The AI & tech Talk Show with Lou, Loïc and Nico.","Du Talk Show au concret":"From Talk Show to projects","Le Talk Show IA et tech de Lou, Loïc et Nico. Chaque semaine, leurs essais d'outils IA et les projets qu'ils construisent avec.":"Lou, Loïc and Nico's AI and tech podcast. Every week, their AI tool experiments and the projects they build with them.","Premier épisode (pilote) de Sandbox, le Talk Show qui teste, débat et remet les chiffres en perspective. Lou, Nicolas et Loïc passent en revue l'actualité IA et tech de la semaine, présentent un outil fait maison et racontent leurs expérimentations en cours.":"First pilot episode of Sandbox, the Talk Show that tests, debates and puts numbers into perspective. Lou, Nicolas and Loïc review the week's AI and tech news, present a home-built tool and discuss their ongoing experiments.","carrousel":"carousel","Projets de l'équipe":"Projects by the team","Projet précédent":"Previous project","Projet suivant":"Next project","Épisodes":"Episodes","Talk Show":"Talk Show","Talk Show · ressources · classements":"Talk Show · resources · rankings","Talk Show IA":"AI Talk Show","Voir la courbe chronologique":"View the timeline chart","Illustrations de concept, sans valeur de document d'archive.":"Concept illustrations, not archival documents.","Les références renvoient aux sources primaires.":"References link to primary sources.","Inventions voisines dans la chronologie":"Nearby inventions in the timeline","Créateur de quêtes du quotidien":"Creator of everyday quests","Inventeur, même sans Wi-Fi":"Inventor, even without Wi-Fi","Chef d'orchestre des agents":"Conductor of AI agents","Faire connaissance":"Get to know us","Parcourir la chronologie":"Browse the timeline","{name} : comprendre l'IA en la pratiquant":"{name}: understanding AI through practice","Dans cet épisode (extraits)":"In this episode (highlights)","Intro : le format Sandbox (Radar, Build, Clash, Bêta)":"Introduction: the Sandbox format (Radar, Build, Clash, Bêta)","Le Radar · Impulsion IA 2027 : 80 M€ pour former à l’IA":"Le Radar · Impulsion IA 2027: €80 million for AI training","Le Radar · Vibe coder l’actualité (Vincent Flibustier)":"Le Radar · Vibe coding the news (Vincent Flibustier)","Le Radar · Google Playground : créer un jeu à partir d’un prompt":"Le Radar · Google Playground: create a game from a prompt","Le Radar · Mistral Large 4, le « chonk »":"Le Radar · Mistral Large 4, the \"chonk\"","Le Radar · Claude Haiku 5.5 : 10× moins cher":"Le Radar · Claude Haiku 5.5: 10× cheaper","Le Radar · Claude Startups : crédits et conditions":"Le Radar · Claude Startups: credits and terms","Le Radar · Gemini at Work 2026 : l’intégration avant le modèle ?":"Le Radar · Gemini at Work 2026: integration before the model?","Le Build · Sillage, une brique au-dessus des harnais (Nicolas)":"Le Build · Sillage, a layer above the harnesses (Nicolas)","Le Clash · Sillage est-il un harnais ? Codex, Claude Code, OpenCode":"Le Clash · Is Sillage a harness? Codex, Claude Code, OpenCode","Le Clash · Contexte, mémoire et « context rot »":"Le Clash · Context, memory and \"context rot\"","Le Clash · Dispatch, orchestration et interfaces":"Le Clash · Dispatch, orchestration and interfaces","La Bêta · Lou : une chaîne YouTube Lo-Fi 100 % IA et des skills":"La Bêta · Lou: a 100% AI Lo-Fi YouTube channel and skills","La Bêta · Loïc : Nemotron 3.5, Strata, Qwen Image 2.1":"La Bêta · Loïc: Nemotron 3.5, Strata, Qwen Image 2.1","La Bêta · Nicolas : jeux vidéo vibe-codés et Power Rush":"La Bêta · Nicolas: vibe-coded video games and Power Rush","Impulsion IA, 12 mesures (ministère du Travail":"Impulsion IA, 12 measures (Ministry of Labour","Émeutes lycées simulator, le jeu vibe-codé de Vincent Flibustier":"Émeutes lycées simulator, the vibe-coded game by Vincent Flibustier","Claude Haiku 5.5 sur Artificial Analysis":"Claude Haiku 5.5 on Artificial Analysis","Avenant startups":"Startup addendum","Conditions commerciales":"Commercial terms","Conditions par service (ZDR":"Service-specific terms (ZDR","Le site et le film de présentation":"The website and introduction video","Le dépôt GitHub (open source, MIT":"The GitHub repository (open source, MIT","Power Rush, le jeu de Nicolas":"Power Rush, the game by Nicolas","Explorez les podcasts vidéo Sandbox sur l’IA, le développement et l’open source. Retrouvez les sources de chaque épisode et nos classements tech hebdomadaires.":"Explore Sandbox video podcasts about AI, development and open source. Find the sources for each episode and our weekly tech rankings.","Quatre séquences : Le Radar (l'actu), Le Build (un outil qu'on a construit), Le Clash (on débat) et La Bêta (nos tests du moment).":"Four segments: Le Radar (the news), Le Build (a tool we built), Le Clash (our debate) and La Bêta (our current experiments).","SOURCES DU RADAR":"RADAR SOURCES","Conditions par service (ZDR)":"Service-specific terms (ZDR)","Quatre séquences : Le Radar (l'actu), Le Build (un outil qu'on a construit), Le Clash (on débat) et La Bêta (nos tests du moment). 🔗 SOURCES DU RADAR • Impulsion IA, 12 mesures (ministère du Travail) : https://travail-emploi.gouv.fr/impulsion-ia-12-mesures-pour-impulser-anticiper-et-dialoguer-face-la-transformation-du-monde-du-travail-par-lintelligence-artificielle • Émeutes lycées simulator, le jeu vibe-codé de Vincent Flibustier : https://emeutes-2eyfvtil.manus.game/ • Google Labs Playground : https://labs.google/playground • Mistral Large 4 : https://mistral.ai/news/mistral-large-4/ • Claude Haiku 5.5 (Anthropic) : https://www.anthropic.com/claude-haiku-5-5 • Claude Haiku 5.5 sur Artificial Analysis : https://artificialanalysis.ai/changelog • Claude Startups : https://claude.com/fr/programs/startups – Avenant startups : https://www.anthropic.com/legal/startup-program-addendum – Conditions commerciales : https://www.anthropic.com/legal/commercial-terms – Conditions par service (ZDR) : https://www.anthropic.com/legal/service-specific-terms • Gemini at Work 2026 (Google Cloud) : https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026 🛠️ LE BUILD : SILLAGE • Le site et le film de présentation : https://marlburrow.github.io/sillage/ • Le dépôt GitHub (open source, MIT) : https://github.com/MarlBurroW/sillage 🧪 LA BÊTA • Power Rush, le jeu de Nicolas : https://powerrush.marlburrow.dev":"$10","Épisode 1":"Episode 1","Conclusion":"Conclusion","Google Labs Playground":"Google Labs Playground","Mistral Large 4":"Mistral Large 4","Claude Haiku 5.5 (Anthropic":"Claude Haiku 5.5 (Anthropic","Claude Startups":"Claude Startups","Gemini at Work 2026 (Google Cloud":"Gemini at Work 2026 (Google Cloud","Mise à jour : 16 octobre 2026":"Updated: 16 October 2026","Vercel Speed Insights mesure aussi les temps de chargement, la réactivité et la stabilité visuelle des pages. Il suit le même choix de consentement. Les paramètres d'URL et la zone d'administration sont exclus de nos envois.":"Vercel Speed Insights also measures page loading times, responsiveness and visual stability. It follows the same consent choice. URL parameters and the administration area are excluded from the data we send.","Nous utilisons Google Analytics, Vercel Analytics et Speed Insights pour mesurer la fréquentation, les actions et la vitesse du site, avec votre accord. Votre choix vaut 180 jours. Vous pouvez le retirer à tout moment.":"With your consent, we use Google Analytics, Vercel Analytics and Speed Insights to measure site traffic, actions and speed. Your choice lasts 180 days. You can withdraw it at any time.","Usamos Google Analytics, Vercel Analytics y Speed Insights para medir el uso y la velocidad del sitio. Estos servicios solo se activan con tu consentimiento. Puedes cambiar tu elección en cualquier momento.":"With your consent, we use Google Analytics, Vercel Analytics and Speed Insights to measure site traffic, actions and speed. Your choice lasts 180 days. You can withdraw it at any time.","We use Google Analytics, Vercel Analytics and Speed Insights to measure site traffic and speed. These services are enabled only with your consent. You can change your choice at any time.":"With your consent, we use Google Analytics, Vercel Analytics and Speed Insights to measure site traffic, actions and speed. Your choice lasts 180 days. You can withdraw it at any time.","Wir verwenden Google Analytics, Vercel Analytics und Speed Insights, um die Nutzung und Geschwindigkeit der Website zu messen. Diese Dienste werden erst nach Ihrer Zustimmung aktiviert. Sie können Ihre Auswahl jederzeit ändern.":"With your consent, we use Google Analytics, Vercel Analytics and Speed Insights to measure site traffic, actions and speed. Your choice lasts 180 days. You can withdraw it at any time.","On a déjà abordé le sujet dans le podcast, et on y revient régulièrement. Les deux projets cités ont leur fiche, avec leur historique de classement.":"We have already discussed this topic on the podcast and return to it regularly. Both projects have their own pages, including their ranking history.","Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.":"Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.","Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. De":"Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.","Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent experience":"Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent experience","Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent":"Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent experience","The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]":"The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]","The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format wit":"The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]","The fastest, cheapest, most efficient open-source agent harness. Run more than 500 agents on a $10 VPS.":"The fastest, cheapest, most efficient open-source agent harness. Run more than 500 agents on a $10 VPS.","Open-source Rust agent engine and terminal client for Codewhale, with provider choice, tools, approvals and receipts.":"Open-source Rust agent engine and terminal client for Codewhale, with provider choice, tools, approvals and receipts.","Open-source Rust agent engine and terminal client for Codewhale, with provider choice, tools, approvals and re":"Open-source Rust agent engine and terminal client for Codewhale, with provider choice, tools, approvals and receipts.","An open-source engine that turns continuous content streams into search and monitoring. Durable ingestion, hybrid search, alerts, and plugins for formats, models and business rules.":"An open-source engine that turns continuous content streams into search and monitoring. Durable ingestion, hybrid search, alerts, and plugins for formats, models and business rules.","An open-source engine that turns continuous content streams into search and monitoring. Durable ingestion, hyb":"An open-source engine that turns continuous content streams into search and monitoring. Durable ingestion, hybrid search, alerts, and plugins for formats, models and business rules.","A reliable coding agent for complex software engineering tasks.":"A reliable coding agent for complex software engineering tasks.","Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with small models for free":"Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with small models for free","Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with":"Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with small models for free","Python scraper based on AI":"Python scraper based on AI","Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.":"Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.","Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to":"Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.","Mastra is the modern TypeScript framework for AI-powered applications and agents.":"Mastra is the modern TypeScript framework for AI-powered applications and agents.","High performance coding agent harness written in rust":"High performance coding agent harness written in rust","Open source agentic operating system":"Open source agentic operating system","💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows":"💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows","Formerly KrillinAI, VideoTranslation and Dubbing Tool. Open-source AI workspace for creators, powered by Codex. Video downloading, video translation, create videos, images, voice, avatars, and edits with Agents in one place. Suitable for your Social Media like YouTube, TikTok, Instagram, etc.":"Formerly KrillinAI, VideoTranslation and Dubbing Tool. Open-source AI workspace for creators, powered by Codex. Video downloading, video translation, create videos, images, voice, avatars, and edits with Agents in one place. Suitable for your Social Media like YouTube, TikTok, Instagram, etc.","Formerly KrillinAI, VideoTranslation and Dubbing Tool. Open-source AI workspace for creators, powered by Codex":"Formerly KrillinAI, VideoTranslation and Dubbing Tool. Open-source AI workspace for creators, powered by Codex. Video downloading, video translation, create videos, images, voice, avatars, and edits with Agents in one place. Suitable for your Social Media like YouTube, TikTok, Instagram, etc.","LMCache: Supercharge Your LLM with the Fastest KV Cache Layer":"LMCache: Supercharge Your LLM with the Fastest KV Cache Layer","Agentic Development Environment based on OpenCode AI agent":"Agentic Development Environment based on OpenCode AI agent","Ekko Studio is a local-first AI workspace for multi-agent chat, coding, and visual workflows, available on desktop and the web.":"Ekko Studio is a local-first AI workspace for multi-agent chat, coding, and visual workflows, available on desktop and the web.","Ekko Studio is a local-first AI workspace for multi-agent chat, coding, and visual workflows, available on des":"Ekko Studio is a local-first AI workspace for multi-agent chat, coding, and visual workflows, available on desktop and the web.","⚙️🦀 Build modular and scalable LLM Applications in Rust":"⚙️🦀 Build modular and scalable LLM Applications in Rust","Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.":"Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.","Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScr":"Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.","Skills for Real Engineers. Straight from my .agents directory.":"Skills for Real Engineers. Straight from my .agents directory.","Official agent plugin providing skills and MCP server configurations for Azure scenarios.":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","CLI for AI-powered web scraping, structured data extraction, web search, multi-page crawling, and scheduled page-change monitoring powered by the ScrapeGraph AI API. Turn any URL into markdown, HTML, screenshots, links, images, summaries, or brand assets; pull structured JSON with a prompt + schema; watch pages on a cron with webhook alert":"CLI for AI-powered web scraping, structured data extraction, web search, multi-page crawling, and scheduled page-change monitoring powered by the ScrapeGraph AI API. Turn any URL into markdown, HTML, screenshots, links, images, summaries, or brand assets; pull structured JSON with a prompt + schema; watch pages on a cron with webhook alert","CLI for AI-powered web scraping, structured data extraction, web search, multi-page crawling, and scheduled":"CLI for AI-powered web scraping, structured data extraction, web search, multi-page crawling, and scheduled page-change monitoring powered by the ScrapeGraph AI API. Turn any URL into markdown, HTML, screenshots, links, images, summaries, or brand assets; pull structured JSON with a prompt + schema; watch pages on a cron with webhook alert","Write HTML. Render video. Built for agents.":"Write HTML. Render video. Built for agents.","Gardez une longueur":"Stay one step","d'avance sur la tech.":"ahead in tech.","Faites défiler pour découvrir l'équipe.":"Scroll to meet the team."},"children":["$L11","$L12","$L13","$L14"]}],"$L15","$L16"]}]}] 17:I[60664,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"Text"] 18:I[81210,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"ScrollState"] 19:I[60664,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"LocalizedLink"] 1a:I[63493,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"Image"] 1b:I[59955,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js"],"MainNav"] 1c:I[59955,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js"],"LocaleSwitcher"] 1d:I[59955,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js"],"MobileNav"] 1e:I[60664,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"LocalizedElement"] 1f:I[98852,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"AnalyticsConsent"] 11:["$","a",null,{"className":"skip","href":"#main","children":["$","$L17",null,{"children":"Aller au contenu"}]}] 12:[["$","$L18",null,{}],["$","header",null,{"className":"site-header border-b border-hair bg-paper","children":["$","div",null,{"className":"wrap masthead-row","children":[["$","$L19",null,{"href":"/","aria-label":"Sandbox — Accueil Sandbox","className":"shrink-0","children":["$","span",null,{"className":"site-brand ","children":["$","span",null,{"className":"wordmark","children":["$","$L1a",null,{"className":"wordmark-image","src":"/sandbox-logo.png","alt":"Sandbox","width":2172,"height":724}]}]}]}],["$","div",null,{"className":"desktop-navigation","children":["$","$L1b",null,{"locale":"en"}]}],["$","div",null,{"className":"masthead-tools","children":[["$","$L19",null,{"href":"/search","className":"site-search-action","aria-label":"Rechercher dans le Talk Show et les classements","children":[["$","svg",null,{"viewBox":"0 0 24 24","width":"20","height":"20","fill":"none","stroke":"currentColor","strokeWidth":"2","aria-hidden":"true","children":[["$","circle",null,{"cx":"10","cy":"10","r":"6"}],["$","path",null,{"d":"m15 15 6 6"}]]}],["$","span",null,{"className":"sr-only","children":["$","$L17",null,{"children":"Rechercher"}]}]]}],["$","$L1c",null,{"locale":"en"}]]}],["$","$L1d",null,{"locale":"en"}]]}]}]] 13:["$","main",null,{"id":"main","tabIndex":-1,"children":["$","$L2",null,{"parallelRouterKey":"children","error":"$undefined","errorStyles":"$undefined","errorScripts":"$undefined","template":["$","$L3",null,{}],"templateStyles":"$undefined","templateScripts":"$undefined","notFound":[["$","div",null,{"className":"wrap not-found-page","children":[["$","p",null,{"className":"eyebrow","children":["$","$L17",null,{"children":"Erreur 404"}]}],["$","h1",null,{"children":["$","$L17",null,{"children":"Cette page est introuvable."}]}],["$","p",null,{"children":["$","$L17",null,{"children":"Le lien a peut-être changé. Retrouvez un épisode, explorez les classements ou lancez une recherche."}]}],["$","div",null,{"className":"page-actions","children":[["$","$L19",null,{"href":"/podcast","className":"btn btn-solid","children":["$","$L17",null,{"children":"Les épisodes"}]}],["$","$L19",null,{"href":"/charts","className":"btn","children":["$","$L17",null,{"children":"Les classements"}]}],["$","$L19",null,{"href":"/search","className":"btn","children":["$","$L17",null,{"children":"Rechercher"}]}]]}]]}],[]],"forbidden":"$undefined","unauthorized":"$undefined"}]}] 14:["$","footer",null,{"className":"site-footer discovery-footer","children":[["$","div",null,{"className":"wrap discovery-footer-grid","children":[["$","div",null,{"children":[["$","span",null,{"className":"site-brand footer-wordmark","children":["$","span",null,{"className":"wordmark","children":["$","$L1a",null,{"className":"wordmark-image","src":"/sandbox-logo.png","alt":"Sandbox","width":2172,"height":724}]}]}],["$","p",null,{"children":["$","$L17",null,{"children":"Le Talk Show IA & tech de Lou, Loïc et Nico."}]}],["$","div",null,{"className":"footer-platforms","children":[["$","a","youtube",{"href":"https://www.youtube.com/@SandboxFR","target":"_blank","rel":"noopener noreferrer","data-analytics-event":"youtube_open","data-analytics-placement":"footer","data-analytics-platform":"youtube","children":[["$","svg",null,{"width":"18","height":"18","viewBox":"0 0 24 24","fill":"currentColor","aria-hidden":"true","children":["$","path",null,{"d":"M23 7a3 3 0 0 0-2-2C19 4.5 12 4.5 12 4.5S5 4.5 3 5a3 3 0 0 0-2 2 31 31 0 0 0 0 10 3 3 0 0 0 2 2c2 .5 9 .5 9 .5s7 0 9-.5a3 3 0 0 0 2-2 31 31 0 0 0 0-10ZM9.5 15.5v-7l6 3.5-6 3.5Z"}]}],"YouTube",["$","span",null,{"className":"sr-only","children":["$","$L17",null,{"children":"(nouvel onglet)"}]}]]}],["$","a","spotify",{"href":"https://open.spotify.com/show/2Wp6S4qgRqQaV01AjkDcMH","target":"_blank","rel":"noopener noreferrer","data-analytics-event":"listen_click","data-analytics-placement":"footer","data-analytics-platform":"spotify","children":[["$","svg",null,{"width":"18","height":"18","viewBox":"0 0 24 24","fill":"currentColor","aria-hidden":"true","children":["$","path",null,{"d":"M12 2a10 10 0 1 0 0 20 10 10 0 0 0 0-20Zm4.6 14.5a.75.75 0 0 1-1 .25c-2.7-1.65-5.65-1.9-9.1-1.1a.75.75 0 1 1-.35-1.45c3.85-.9 7.25-.6 10.2 1.25a.75.75 0 0 1 .25 1.05Zm1.2-2.7a.9.9 0 0 1-1.25.3c-3.05-1.85-7.65-2.4-11.2-1.3a.9.9 0 1 1-.55-1.7c4.05-1.25 9.15-.65 12.7 1.5a.9.9 0 0 1 .3 1.2Zm.1-2.8C14.25 8.85 8.2 8.65 4.7 9.7a1.1 1.1 0 1 1-.65-2.1c4.05-1.2 10.75-.95 14.95 1.5a1.1 1.1 0 0 1-1.1 1.9Z"}]}],"Spotify",["$","span",null,{"className":"sr-only","children":["$","$L17",null,{"children":"(nouvel onglet)"}]}]]}]]}]]}],["$","$L1e",null,{"as":"nav","aria-label":"Explorer","children":[["$","h2",null,{"children":["$","$L17",null,{"children":"Explorer"}]}],["$","ul",null,{"children":[["$","li",null,{"children":["$","$L19",null,{"href":"/charts","children":["$","$L17",null,{"children":"Classements"}]}]}],["$","li",null,{"children":["$","$L19",null,{"href":"/time-machine","children":["$","$L17",null,{"children":"Time Machine (en français)"}]}]}],["$","li",null,{"children":["$","$L19",null,{"href":"/nos-projets","children":["$","$L17",null,{"children":"Nos projets"}]}]}],["$","li",null,{"children":["$","$L19",null,{"href":"/search","children":["$","$L17",null,{"children":"Rechercher"}]}]}]]}]]}],["$","$L1e",null,{"as":"nav","aria-label":"Sandbox","children":[["$","h2",null,{"children":"Sandbox"}],["$","ul",null,{"children":[["$","li",null,{"children":["$","$L19",null,{"href":"/podcast","children":["$","$L17",null,{"children":"Le Talk Show"}]}]}],["$","li",null,{"children":["$","$L19",null,{"href":"/about","children":["$","$L17",null,{"children":"À propos"}]}]}],["$","li",null,{"children":["$","$L19",null,{"href":"/charts/history","children":["$","$L17",null,{"children":"Archives des classements"}]}]}],["$","li",null,{"children":["$","$L19",null,{"href":"/charts/awards","children":["$","$L17",null,{"children":"Badges & Awards"}]}]}]]}]]}]]}],["$","div",null,{"className":"wrap discovery-footer-bottom","children":[["$","span",null,{"children":["© ",2026," SANDBOX"]}],["$","a",null,{"href":"mailto:sandbox.podcast.pro@gmail.com","children":"sandbox.podcast.pro@gmail.com"}],["$","$L19",null,{"href":"/confidentialite","children":["$","$L17",null,{"children":"Données personnelles"}]}],["$","a",null,{"href":"/feed.xml","children":["$","$L17",null,{"children":"Flux RSS"}]}]]}]]}] 15:["$","script",null,{"type":"application/ld+json","dangerouslySetInnerHTML":{"__html":"[{\"@context\":\"https://schema.org\",\"@type\":\"WebSite\",\"name\":\"Sandbox\",\"url\":\"https://www.sandboxpodcast.fr\",\"description\":\"Explore Sandbox video podcasts about AI, development and open source. Find the sources for each episode and our weekly tech rankings.\",\"inLanguage\":\"en\"},{\"@context\":\"https://schema.org\",\"@type\":\"Organization\",\"name\":\"Sandbox\",\"url\":\"https://www.sandboxpodcast.fr\",\"logo\":\"https://www.sandboxpodcast.fr/sandbox-logo.png\"}]"}}] 16:["$","$L1f",null,{"language":"en"}] 20:I[7235,["/_next/static/immutable/chunks/3oguwynlpfg8i.js","/_next/static/immutable/chunks/0xzowx1_cz2pp.js","/_next/static/immutable/chunks/06gfbzk5l_5yy.js","/_next/static/immutable/chunks/24mhef7zkfs6-.js","/_next/static/immutable/chunks/2ghckoypcz7ea.js","/_next/static/immutable/chunks/07a5qibpo8z0n.js","/_next/static/immutable/chunks/3iesc6gvkoa3o.js","/_next/static/immutable/chunks/1maqrd5tp0foz.js"],"ChartsExperience"] 21:I[14614,["/_next/static/immutable/chunks/357cxq-i-o0b8.js"],"IconMark"] 5:["$","$L20",null,{"data":{"checkedAt":"2026-10-16T21:18:33.182Z","mode":"live","week":"2026-W41","weeks":["2026-10-12","2026-10-11","2026-10-10","2026-W41"],"entities":[{"slug":"1jehuang-jcode-1128204614","name":"jcode","kind":"project","org":"1jehuang","tagline":"High performance coding agent harness written in rust","description":"High performance coding agent harness written in rust","category":"Other","topics":[],"links":[{"kind":"github","label":"1jehuang/jcode","url":"https://github.com/1jehuang/jcode"}],"href":"/charts/project/1jehuang-jcode-1128204614","github":{"repo":"1jehuang/jcode"}},{"slug":"esengine-deepseek-reasonix-1216785679","name":"DeepSeek-Reasonix","kind":"project","org":"esengine","tagline":"A reliable coding agent for complex software engineering tasks.","description":"A reliable coding agent for complex software engineering tasks.","category":"Other","topics":[],"links":[{"kind":"github","label":"esengine/DeepSeek-Reasonix","url":"https://github.com/esengine/DeepSeek-Reasonix"}],"href":"/charts/project/esengine-deepseek-reasonix-1216785679","github":{"repo":"esengine/DeepSeek-Reasonix"}},{"slug":"claudeopus55high-arena-anthropic-claude-opus-5-5-high","name":"claude-opus-5.5-high","kind":"model","org":"anthropic","tagline":"claude-opus-5.5-high","description":"claude-opus-5.5-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudeopus55high-arena-anthropic-claude-opus-5-5-high"},{"slug":"qwen38max0902","name":"qwen3.8-max-0902","kind":"model","org":"alibaba","tagline":"qwen3.8-max-0902","description":"qwen3.8-max-0902","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"},{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/qwen38max0902"},{"slug":"0xplaygrounds-rig-810861466","name":"rig","kind":"project","org":"0xPlaygrounds","tagline":"⚙️🦀 Build modular and scalable LLM Applications in Rust","description":"⚙️🦀 Build modular and scalable LLM Applications in Rust","category":"Other","topics":[],"links":[{"kind":"github","label":"0xPlaygrounds/rig","url":"https://github.com/0xPlaygrounds/rig"}],"href":"/charts/project/0xplaygrounds-rig-810861466","github":{"repo":"0xPlaygrounds/rig"}},{"slug":"neuml-txtai-286301447","name":"txtai","kind":"project","org":"neuml","tagline":"💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows","description":"💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows","category":"Other","topics":[],"links":[{"kind":"github","label":"neuml/txtai","url":"https://github.com/neuml/txtai"}],"href":"/charts/project/neuml-txtai-286301447","github":{"repo":"neuml/txtai"}},{"slug":"mastra-ai-mastra-839037098","name":"mastra","kind":"project","org":"mastra-ai","tagline":"Mastra is the modern TypeScript framework for AI-powered applications and agents.","description":"Mastra is the modern TypeScript framework for AI-powered applications and agents.","category":"Other","topics":[],"links":[{"kind":"github","label":"mastra-ai/mastra","url":"https://github.com/mastra-ai/mastra"}],"href":"/charts/project/mastra-ai-mastra-839037098","github":{"repo":"mastra-ai/mastra"}},{"slug":"elizaos-eliza-826170402","name":"eliza","kind":"project","org":"elizaOS","tagline":"Open source agentic operating system","description":"Open source agentic operating system","category":"Other","topics":[],"links":[{"kind":"github","label":"elizaOS/eliza","url":"https://github.com/elizaOS/eliza"}],"href":"/charts/project/elizaos-eliza-826170402","github":{"repo":"elizaOS/eliza"}},{"slug":"scrapegraphai-scrapegraph-ai-749126547","name":"Scrapegraph-ai","kind":"project","org":"ScrapeGraphAI","tagline":"Python scraper based on AI","description":"Python scraper based on AI","category":"Other","topics":[],"links":[{"kind":"github","label":"ScrapeGraphAI/Scrapegraph-ai","url":"https://github.com/ScrapeGraphAI/Scrapegraph-ai"}],"href":"/charts/project/scrapegraphai-scrapegraph-ai-749126547","github":{"repo":"ScrapeGraphAI/Scrapegraph-ai"}},{"slug":"ekkolearnai-ekko-studio-1207639951","name":"ekko-studio","kind":"project","org":"EKKOLearnAI","tagline":"Ekko Studio is a local-first AI workspace for multi-agent chat, coding, and visual workflows, available on des","description":"Ekko Studio is a local-first AI workspace for multi-agent chat, coding, and visual workflows, available on desktop and the web.","category":"Other","topics":[],"links":[{"kind":"github","label":"EKKOLearnAI/ekko-studio","url":"https://github.com/EKKOLearnAI/ekko-studio"}],"href":"/charts/project/ekkolearnai-ekko-studio-1207639951","github":{"repo":"EKKOLearnAI/ekko-studio"}},{"slug":"krillinai-opencreator-904490600","name":"OpenCreator","kind":"project","org":"krillinai","tagline":"Formerly KrillinAI, VideoTranslation and Dubbing Tool. Open-source AI workspace for creators, powered by Codex","description":"Formerly KrillinAI, VideoTranslation and Dubbing Tool. Open-source AI workspace for creators, powered by Codex. Video downloading, video translation, create videos, images, voice, avatars, and edits with Agents in one place. Suitable for your Social Media like YouTube, TikTok, Instagram, etc.","category":"Other","topics":[],"links":[{"kind":"github","label":"krillinai/OpenCreator","url":"https://github.com/krillinai/OpenCreator"}],"href":"/charts/project/krillinai-opencreator-904490600","github":{"repo":"krillinai/OpenCreator"}},{"slug":"mintplex-labs-anything-llm-649170660","name":"anything-llm","kind":"project","org":"Mintplex-Labs","tagline":"Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent ","description":"Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent experience ","category":"Other","topics":[],"links":[{"kind":"github","label":"Mintplex-Labs/anything-llm","url":"https://github.com/Mintplex-Labs/anything-llm"}],"href":"/charts/project/mintplex-labs-anything-llm-649170660","github":{"repo":"Mintplex-Labs/anything-llm"}},{"slug":"skill-mattpocock-skills-to-issues-00543cd7","name":"to-issues","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/to-issues"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-to-issues-00543cd7"},{"slug":"skill-mattpocock-skills-writing-great-skills-dbb203e0","name":"writing-great-skills","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/writing-great-skills"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-writing-great-skills-dbb203e0"},{"slug":"skill-mattpocock-skills-diagnose-74203c86","name":"diagnose","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/diagnose"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-diagnose-74203c86"},{"slug":"skill-mattpocock-skills-design-an-interface-c7e98cca","name":"design-an-interface","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/design-an-interface"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-design-an-interface-c7e98cca"},{"slug":"tinyhumansai-openhuman-1161163182","name":"openhuman","kind":"project","org":"tinyhumansai","tagline":"The fastest, cheapest, most efficient open-source agent harness. Run more than 500 agents on a $10 VPS.","description":"The fastest, cheapest, most efficient open-source agent harness. Run more than 500 agents on a $10 VPS.","category":"Other","topics":[],"links":[{"kind":"github","label":"tinyhumansai/openhuman","url":"https://github.com/tinyhumansai/openhuman"}],"href":"/charts/project/tinyhumansai-openhuman-1161163182","github":{"repo":"tinyhumansai/openhuman"}},{"slug":"composiohq-composio-762304524","name":"composio","kind":"project","org":"ComposioHQ","tagline":"Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to ","description":"Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.","category":"Other","topics":[],"links":[{"kind":"github","label":"ComposioHQ/composio","url":"https://github.com/ComposioHQ/composio"}],"href":"/charts/project/composiohq-composio-762304524","github":{"repo":"ComposioHQ/composio"}},{"slug":"topoteretes-cognee-679343504","name":"cognee","kind":"project","org":"topoteretes","tagline":"Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with ","description":"Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with small models for free","category":"Other","topics":[],"links":[{"kind":"other","label":"OpenSSF Scorecard","url":"https://scorecard.dev/viewer/?uri=github.com/topoteretes/cognee"}],"href":"/charts/project/topoteretes-cognee-679343504","github":{"repo":"topoteretes/cognee"}},{"slug":"claudeopus5high-arena-anthropic-claude-opus-5-high","name":"claude-opus-5-high","kind":"model","org":"anthropic","tagline":"claude-opus-5-high","description":"claude-opus-5-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudeopus5high-arena-anthropic-claude-opus-5-high"},{"slug":"musespark","name":"muse-spark","kind":"model","org":"meta","tagline":"muse-spark","description":"muse-spark","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/musespark"},{"slug":"berriai-litellm-671269505","name":"litellm","kind":"project","org":"BerriAI","tagline":"The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format wit","description":"The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]","category":"Other","topics":[],"links":[{"kind":"other","label":"OpenSSF Scorecard","url":"https://scorecard.dev/viewer/?uri=github.com/BerriAI/litellm"}],"href":"/charts/project/berriai-litellm-671269505","github":{"repo":"BerriAI/litellm"}},{"slug":"openchamber-openchamber-1054790989","name":"openchamber","kind":"project","org":"openchamber","tagline":"Agentic Development Environment based on OpenCode AI agent","description":"Agentic Development Environment based on OpenCode AI agent","category":"Other","topics":[],"links":[{"kind":"github","label":"openchamber/openchamber","url":"https://github.com/openchamber/openchamber"}],"href":"/charts/project/openchamber-openchamber-1054790989","github":{"repo":"openchamber/openchamber"}},{"slug":"codewhale-hq-codewhale-1137711311","name":"Codewhale","kind":"project","org":"codewhale-hq","tagline":"Open-source Rust agent engine and terminal client for Codewhale, with provider choice, tools, approvals and re","description":"Open-source Rust agent engine and terminal client for Codewhale, with provider choice, tools, approvals and receipts.","category":"Other","topics":[],"links":[{"kind":"github","label":"codewhale-hq/Codewhale","url":"https://github.com/codewhale-hq/Codewhale"}],"href":"/charts/project/codewhale-hq-codewhale-1137711311","github":{"repo":"codewhale-hq/Codewhale"}},{"slug":"langgenius-dify-626805178","name":"dify","kind":"project","org":"langgenius","tagline":"Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. De","description":"Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.","category":"Other","topics":[],"links":[{"kind":"github","label":"langgenius/dify","url":"https://github.com/langgenius/dify"}],"href":"/charts/project/langgenius-dify-626805178","github":{"repo":"langgenius/dify"}},{"slug":"gemini4argonhigh-arena-google-gemini-4-argon-high","name":"gemini-4-argon-high","kind":"model","org":"google","tagline":"gemini-4-argon-high","description":"gemini-4-argon-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/gemini4argonhigh-arena-google-gemini-4-argon-high"},{"slug":"claudefable5high-arena-anthropic-claude-fable-5-high","name":"claude-fable-5-high","kind":"model","org":"anthropic","tagline":"claude-fable-5-high","description":"claude-fable-5-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudefable5high-arena-anthropic-claude-fable-5-high"},{"slug":"claudeopus46high","name":"claude-opus-4-6-high","kind":"model","org":"anthropic","tagline":"claude-opus-4-6-high","description":"claude-opus-4-6-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudeopus46high"},{"slug":"claudeopus47high","name":"claude-opus-4-7-high","kind":"model","org":"anthropic","tagline":"claude-opus-4-7-high","description":"claude-opus-4-7-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudeopus47high"},{"slug":"claudefable51max-arena-anthropic-claude-fable-5-1-max","name":"claude-fable-5.1-max","kind":"model","org":"anthropic","tagline":"claude-fable-5.1-max","description":"claude-fable-5.1-max","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudefable51max-arena-anthropic-claude-fable-5-1-max"},{"slug":"claudeopus46","name":"claude-opus-4-6","kind":"model","org":"anthropic","tagline":"claude-opus-4-6","description":"claude-opus-4-6","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudeopus46"},{"slug":"gemini38flashhigh-arena-google-gemini-3-8-flash-high","name":"gemini-3.8-flash-high","kind":"model","org":"google","tagline":"gemini-3.8-flash-high","description":"gemini-3.8-flash-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/gemini38flashhigh-arena-google-gemini-3-8-flash-high"},{"slug":"musespark12xhigh-arena-meta-muse-spark-1-2-xhigh","name":"muse-spark-1.2 (xHigh)","kind":"model","org":"meta","tagline":"muse-spark-1.2 (xHigh)","description":"muse-spark-1.2 (xHigh)","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/musespark12xhigh-arena-meta-muse-spark-1-2-xhigh"},{"slug":"musespark13max-arena-meta-muse-spark-1-3-max","name":"muse-spark-1.3-max","kind":"model","org":"meta","tagline":"muse-spark-1.3-max","description":"muse-spark-1.3-max","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/musespark13max-arena-meta-muse-spark-1-3-max"},{"slug":"musespark13max-artificial-analysis-meta-muse-spark-1-3-20260902","name":"Muse Spark 1.3 (Max)","kind":"model","org":"meta","tagline":"Muse Spark 1.3 (Max)","description":"Muse Spark 1.3 (Max)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/musespark13max-artificial-analysis-meta-muse-spark-1-3-20260902"},{"slug":"claudeopus47","name":"claude-opus-4-7","kind":"model","org":"anthropic","tagline":"claude-opus-4-7","description":"claude-opus-4-7","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudeopus47"},{"slug":"musespark11-arena-meta-muse-spark-1-1","name":"muse-spark-1.1","kind":"model","org":"meta","tagline":"muse-spark-1.1","description":"muse-spark-1.1","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/musespark11-arena-meta-muse-spark-1-1"},{"slug":"claudeopus5max-arena-anthropic-claude-opus-5-max","name":"claude-opus-5-max","kind":"model","org":"anthropic","tagline":"claude-opus-5-max","description":"claude-opus-5-max","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/claudeopus5max-arena-anthropic-claude-opus-5-max"},{"slug":"claudeopus5max-artificial-analysis-anthropic-claude-opus-5-20260723","name":"Claude Opus 5 (Max)","kind":"model","org":"anthropic","tagline":"Claude Opus 5 (Max)","description":"Claude Opus 5 (Max)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/claudeopus5max-artificial-analysis-anthropic-claude-opus-5-20260723"},{"slug":"kimik3max-arena-moonshot-kimi-k3-max","name":"kimi-k3-max","kind":"model","org":"moonshot","tagline":"kimi-k3-max","description":"kimi-k3-max","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/kimik3max-arena-moonshot-kimi-k3-max"},{"slug":"gemini37flashhigh-arena-google-gemini-3-7-flash-high","name":"gemini-3.7-flash-high","kind":"model","org":"google","tagline":"gemini-3.7-flash-high","description":"gemini-3.7-flash-high","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/gemini37flashhigh-arena-google-gemini-3-7-flash-high"},{"slug":"gemini31propreview-arena-google-gemini-3-1-pro-preview","name":"gemini-3.1-pro-preview","kind":"model","org":"google","tagline":"gemini-3.1-pro-preview","description":"gemini-3.1-pro-preview","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/gemini31propreview-arena-google-gemini-3-1-pro-preview"},{"slug":"gemini3pro","name":"gemini-3-pro","kind":"model","org":"google","tagline":"gemini-3-pro","description":"gemini-3-pro","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/gemini3pro"},{"slug":"gpt56solxhigh-arena-openai-gpt-5-6-sol-xhigh","name":"gpt-5.6-sol-xhigh","kind":"model","org":"openai","tagline":"gpt-5.6-sol-xhigh","description":"gpt-5.6-sol-xhigh","category":"Models","topics":[],"links":[{"kind":"other","label":"Arena via Hugging Face","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"}],"href":"/charts/entity/gpt56solxhigh-arena-openai-gpt-5-6-sol-xhigh"},{"slug":"gpt61solmax-artificial-analysis-openai-gpt-6-1-sol-20260929","name":"GPT-6.1 Sol (Max)","kind":"model","org":"openai","tagline":"GPT-6.1 Sol (Max)","description":"GPT-6.1 Sol (Max)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/gpt61solmax-artificial-analysis-openai-gpt-6-1-sol-20260929"},{"slug":"qwen38max-artificial-analysis-qwen-qwen3-8-max-20260803","name":"Qwen3.8 Max","kind":"model","org":"qwen","tagline":"Qwen3.8 Max","description":"Qwen3.8 Max","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/qwen38max-artificial-analysis-qwen-qwen3-8-max-20260803"},{"slug":"mimov26pro-artificial-analysis-xiaomi-mimo-v2-6-pro-20260921","name":"MiMo-V2.6-Pro","kind":"model","org":"xiaomi","tagline":"MiMo-V2.6-Pro","description":"MiMo-V2.6-Pro","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/mimov26pro-artificial-analysis-xiaomi-mimo-v2-6-pro-20260921"},{"slug":"glm53max-artificial-analysis-z-ai-glm-5-3-20260816","name":"GLM-5.3 (Max)","kind":"model","org":"z-ai","tagline":"GLM-5.3 (Max)","description":"GLM-5.3 (Max)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/glm53max-artificial-analysis-z-ai-glm-5-3-20260816"},{"slug":"gpt6astramax-artificial-analysis-openai-gpt-6-astra-20260903","name":"GPT-6 Astra (Max)","kind":"model","org":"openai","tagline":"GPT-6 Astra (Max)","description":"GPT-6 Astra (Max)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/gpt6astramax-artificial-analysis-openai-gpt-6-astra-20260903"},{"slug":"gpt6solmax-artificial-analysis-openai-gpt-6-sol-20260922","name":"GPT-6 Sol (Max)","kind":"model","org":"openai","tagline":"GPT-6 Sol (Max)","description":"GPT-6 Sol (Max)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/gpt6solmax-artificial-analysis-openai-gpt-6-sol-20260922"},{"slug":"grok47xhigh-artificial-analysis-x-ai-grok-4-7-20260916","name":"Grok 4.7 (Xhigh)","kind":"model","org":"x-ai","tagline":"Grok 4.7 (Xhigh)","description":"Grok 4.7 (Xhigh)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/grok47xhigh-artificial-analysis-x-ai-grok-4-7-20260916"},{"slug":"skill-mattpocock-skills-to-prd-a1ff2082","name":"to-prd","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/to-prd"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-to-prd-a1ff2082"},{"slug":"claudeopus55maxdefaultfallback","name":"Claude Opus 5.5 (Max, Default Fallback)","kind":"model","org":"anthropic","tagline":"Claude Opus 5.5 (Max, Default Fallback)","description":"Claude Opus 5.5 (Max, Default Fallback)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/claudeopus55maxdefaultfallback"},{"slug":"claudesonnet55maxdefaultfallback","name":"Claude Sonnet 5.5 (Max, Default Fallback)","kind":"model","org":"anthropic","tagline":"Claude Sonnet 5.5 (Max, Default Fallback)","description":"Claude Sonnet 5.5 (Max, Default Fallback)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/claudesonnet55maxdefaultfallback"},{"slug":"claudefable51maxdefaultfallback","name":"Claude Fable 5.1 (Max, Default Fallback)","kind":"model","org":"anthropic","tagline":"Claude Fable 5.1 (Max, Default Fallback)","description":"Claude Fable 5.1 (Max, Default Fallback)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/claudefable51maxdefaultfallback"},{"slug":"claudefable5maxopus48fallback","name":"Claude Fable 5 (Max, Opus 4.8 Fallback)","kind":"model","org":"anthropic","tagline":"Claude Fable 5 (Max, Opus 4.8 Fallback)","description":"Claude Fable 5 (Max, Opus 4.8 Fallback)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/claudefable5maxopus48fallback"},{"slug":"gpt56solmax","name":"GPT-5.6 Sol (Max)","kind":"model","org":"openai","tagline":"GPT-5.6 Sol (Max)","description":"GPT-5.6 Sol (Max)","category":"Models","topics":[],"links":[{"kind":"other","label":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"}],"href":"/charts/entity/gpt56solmax"},{"slug":"the-vibe-company-quivr-640079149","name":"quivr","kind":"project","org":"The-Vibe-Company","tagline":"An open-source engine that turns continuous content streams into search and monitoring. Durable ingestion, hyb","description":"An open-source engine that turns continuous content streams into search and monitoring. Durable ingestion, hybrid search, alerts, and plugins for formats, models and business rules.","category":"Other","topics":[],"links":[{"kind":"github","label":"The-Vibe-Company/quivr","url":"https://github.com/The-Vibe-Company/quivr"}],"href":"/charts/project/the-vibe-company-quivr-640079149","github":{"repo":"The-Vibe-Company/quivr"}},{"slug":"strands-agents-harness-sdk-983715534","name":"harness-sdk","kind":"project","org":"strands-agents","tagline":"Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScr","description":"Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.","category":"Other","topics":[],"links":[{"kind":"github","label":"strands-agents/harness-sdk","url":"https://github.com/strands-agents/harness-sdk"}],"href":"/charts/project/strands-agents-harness-sdk-983715534","github":{"repo":"strands-agents/harness-sdk"}},{"slug":"skill-microsoft-azure-skills-azure-hosted-copilot-sdk-2cb4c2ac","name":"azure-hosted-copilot-sdk","kind":"tool","org":"microsoft","tagline":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","description":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/microsoft/azure-skills/azure-hosted-copilot-sdk"},{"kind":"github","label":"GitHub","url":"https://github.com/microsoft/azure-skills"}],"href":"/charts/entity/skill-microsoft-azure-skills-azure-hosted-copilot-sdk-2cb4c2ac"},{"slug":"skill-mattpocock-skills-edit-article-7b51c229","name":"edit-article","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/edit-article"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-edit-article-7b51c229"},{"slug":"lmcache-lmcache-807305060","name":"LMCache","kind":"project","org":"LMCache","tagline":"LMCache: Supercharge Your LLM with the Fastest KV Cache Layer","description":"LMCache: Supercharge Your LLM with the Fastest KV Cache Layer","category":"Other","topics":[],"links":[{"kind":"other","label":"OpenSSF Scorecard","url":"https://scorecard.dev/viewer/?uri=github.com/LMCache/LMCache"}],"href":"/charts/project/lmcache-lmcache-807305060","github":{"repo":"LMCache/LMCache"}},{"slug":"skill-mattpocock-skills-resolving-merge-conflicts-6366401d","name":"resolving-merge-conflicts","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/resolving-merge-conflicts"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-resolving-merge-conflicts-6366401d"},{"slug":"skill-microsoft-azure-skills-azure-rbac-e2b91e75","name":"azure-rbac","kind":"tool","org":"microsoft","tagline":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","description":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/microsoft/azure-skills/azure-rbac"},{"kind":"github","label":"GitHub","url":"https://github.com/microsoft/azure-skills"}],"href":"/charts/entity/skill-microsoft-azure-skills-azure-rbac-e2b91e75"},{"slug":"skill-microsoft-azure-skills-azure-cost-e8751621","name":"azure-cost","kind":"tool","org":"microsoft","tagline":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","description":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/microsoft/azure-skills/azure-cost"},{"kind":"github","label":"GitHub","url":"https://github.com/microsoft/azure-skills"}],"href":"/charts/entity/skill-microsoft-azure-skills-azure-cost-e8751621"},{"slug":"skill-scrapegraphai-just-scrape-just-scrape-59d9ef84","name":"just-scrape","kind":"tool","org":"ScrapeGraphAI","tagline":"CLI for AI-powered web scraping, structured data extraction, web search, multi-page crawling, and scheduled ","description":"CLI for AI-powered web scraping, structured data extraction, web search, multi-page crawling, and scheduled page-change monitoring powered by the ScrapeGraph AI API. Turn any URL into markdown, HTML, screenshots, links, images, summaries, or brand assets; pull structured JSON with a prompt + schema; watch pages on a cron with webhook alert","category":"Coding","topics":[],"links":[{"kind":"github","label":"GitHub","url":"https://github.com/ScrapeGraphAI/just-scrape"},{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/scrapegraphai/just-scrape/just-scrape"}],"href":"/charts/entity/skill-scrapegraphai-just-scrape-just-scrape-59d9ef84"},{"slug":"skill-mattpocock-skills-write-a-skill-aea77044","name":"write-a-skill","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/write-a-skill"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-write-a-skill-aea77044"},{"slug":"skill-mattpocock-skills-zoom-out-2a509ca1","name":"zoom-out","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/zoom-out"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-zoom-out-2a509ca1"},{"slug":"skill-mattpocock-skills-caveman-09c5ab10","name":"caveman","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/caveman"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-caveman-09c5ab10"},{"slug":"skill-mattpocock-skills-request-refactor-plan-693928dd","name":"request-refactor-plan","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/request-refactor-plan"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-request-refactor-plan-693928dd"},{"slug":"skill-mattpocock-skills-qa-2844017a","name":"qa","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/qa"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-qa-2844017a"},{"slug":"skill-mattpocock-skills-ubiquitous-language-2a562f12","name":"ubiquitous-language","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/ubiquitous-language"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-ubiquitous-language-2a562f12"},{"slug":"skill-mattpocock-skills-obsidian-vault-d4c5f408","name":"obsidian-vault","kind":"tool","org":"mattpocock","tagline":"Skills for Real Engineers. Straight from my .agents directory.","description":"Skills for Real Engineers. Straight from my .agents directory.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/mattpocock/skills/obsidian-vault"},{"kind":"github","label":"GitHub","url":"https://github.com/mattpocock/skills"}],"href":"/charts/entity/skill-mattpocock-skills-obsidian-vault-d4c5f408"},{"slug":"skill-microsoft-azure-skills-azure-cost-optimization-3845cbb2","name":"azure-cost-optimization","kind":"tool","org":"microsoft","tagline":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","description":"Official agent plugin providing skills and MCP server configurations for Azure scenarios.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/microsoft/azure-skills/azure-cost-optimization"},{"kind":"github","label":"GitHub","url":"https://github.com/microsoft/azure-skills"}],"href":"/charts/entity/skill-microsoft-azure-skills-azure-cost-optimization-3845cbb2"},{"slug":"skill-heygen-com-hyperframes-hyperframes-media-831e7348","name":"hyperframes-media","kind":"tool","org":"heygen-com","tagline":"Write HTML. Render video. Built for agents.","description":"Write HTML. Render video. Built for agents.","category":"Productivity","topics":[],"links":[{"kind":"other","label":"Skills.sh","url":"https://www.skills.sh/heygen-com/hyperframes/hyperframes-media"},{"kind":"github","label":"GitHub","url":"https://github.com/heygen-com/hyperframes"}],"href":"/charts/entity/skill-heygen-com-hyperframes-hyperframes-media-831e7348"}],"series":[{"id":"github","slug":"github","primary":"momentum","dimensions":[{"id":"momentum","label":"Dynamique"},{"id":"starVelocity","label":"Vitesse des étoiles"},{"id":"relativeGrowth","label":"Croissance relative"},{"id":"forkVelocity","label":"Vitesse des forks"},{"id":"contributorActivity","label":"Activité des contributeurs"},{"id":"repositoryActivity","label":"Activité du dépôt"}],"metrics":[{"key":"stars","label":"Étoiles","unit":"count","source":"GitHub API","url":"https://docs.github.com/en/graphql/reference/repos"},{"key":"forks","label":"Forks","unit":"count","source":"GitHub API","url":"https://docs.github.com/en/graphql/reference/repos"},{"key":"contributors","label":"Contributeurs","unit":"count","source":"GitHub API","url":"https://docs.github.com/en/graphql/reference/repos"},{"key":"stars7d","label":"Étoiles gagnées (7 jours)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/github/methodology"},{"key":"growth","label":"Croissance (%)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/github/methodology"},{"key":"forks7d","label":"Forks gagnés (7 jours)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/github/methodology"},{"key":"contributorsDelta","label":"Évolution des contributeurs","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/github/methodology"},{"key":"activityScore","label":"Activité du dépôt","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/github/methodology"},{"key":"acceleration","label":"Accélération (%)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/github/methodology"}],"snapshots":[{"chart":"github","week":"2026-10-10","publishedAt":"2026-10-10T09:57:38.967Z","retrievedAt":"2026-10-10T02:35:31.561Z","provenance":"auto","entries":[{"entity":"langgenius-dify-626805178","rank":1,"score":100,"dimensions":{"reach":100,"momentum":100},"metrics":{"forks":24948,"stars":158029,"activityScore":99.97}},{"entity":"mintplex-labs-anything-llm-649170660","rank":2,"score":98.81,"dimensions":{"reach":98.81,"momentum":98.81},"metrics":{"forks":7461,"stars":66870,"activityScore":98.45}},{"entity":"berriai-litellm-671269505","rank":3,"score":97.62,"dimensions":{"reach":97.62,"momentum":97.62},"metrics":{"forks":12193,"stars":60697,"activityScore":100}},{"entity":"tinyhumansai-openhuman-1161163182","rank":4,"score":96.43,"dimensions":{"reach":96.43,"momentum":96.43},"metrics":{"forks":4108,"stars":41751,"activityScore":99.99}},{"entity":"codewhale-hq-codewhale-1137711311","rank":5,"score":95.24,"dimensions":{"reach":95.24,"momentum":95.24},"metrics":{"forks":3566,"stars":41074,"activityScore":99.99}},{"entity":"the-vibe-company-quivr-640079149","rank":6,"score":94.05,"dimensions":{"reach":94.05,"momentum":94.05},"metrics":{"forks":3738,"stars":39577,"activityScore":99.97}},{"entity":"esengine-deepseek-reasonix-1216785679","rank":7,"score":92.86,"dimensions":{"reach":92.86,"momentum":92.86},"metrics":{"forks":2439,"stars":35748,"activityScore":99.99}},{"entity":"topoteretes-cognee-679343504","rank":8,"score":91.67,"dimensions":{"reach":91.67,"momentum":91.67},"metrics":{"forks":3295,"stars":31912,"activityScore":99.44}},{"entity":"scrapegraphai-scrapegraph-ai-749126547","rank":9,"score":90.48,"dimensions":{"reach":90.48,"momentum":90.48},"metrics":{"forks":3229,"stars":31660,"activityScore":88.24}},{"entity":"composiohq-composio-762304524","rank":10,"score":89.29,"dimensions":{"reach":89.29,"momentum":89.29},"metrics":{"forks":4846,"stars":30481,"activityScore":99.8}},{"entity":"mastra-ai-mastra-839037098","rank":11,"score":88.1,"dimensions":{"reach":88.1,"momentum":88.1},"metrics":{"forks":2947,"stars":28673,"activityScore":99.99}},{"entity":"1jehuang-jcode-1128204614","rank":12,"score":86.9,"dimensions":{"reach":86.9,"momentum":86.9},"metrics":{"forks":2370,"stars":20373,"activityScore":99.55}},{"entity":"elizaos-eliza-826170402","rank":13,"score":85.71,"dimensions":{"reach":85.71,"momentum":85.71},"metrics":{"forks":5787,"stars":19561,"activityScore":100}},{"entity":"neuml-txtai-286301447","rank":14,"score":84.52,"dimensions":{"reach":84.52,"momentum":84.52},"metrics":{"forks":927,"stars":13002,"activityScore":98.35}},{"entity":"krillinai-opencreator-904490600","rank":15,"score":83.33,"dimensions":{"reach":83.33,"momentum":83.33},"metrics":{"forks":1306,"stars":12670,"activityScore":99.95}},{"entity":"lmcache-lmcache-807305060","rank":16,"score":82.14,"dimensions":{"reach":82.14,"momentum":82.14},"metrics":{"forks":1994,"stars":11988,"activityScore":99.95}},{"entity":"openchamber-openchamber-1054790989","rank":17,"score":80.95,"dimensions":{"reach":80.95,"momentum":80.95},"metrics":{"forks":1255,"stars":11353,"activityScore":99.8}},{"entity":"ekkolearnai-ekko-studio-1207639951","rank":18,"score":79.76,"dimensions":{"reach":79.76,"momentum":79.76},"metrics":{"forks":1393,"stars":11338,"activityScore":99.95}},{"entity":"0xplaygrounds-rig-810861466","rank":19,"score":78.57,"dimensions":{"reach":78.57,"momentum":78.57},"metrics":{"forks":987,"stars":8840,"activityScore":99.98}},{"entity":"strands-agents-harness-sdk-983715534","rank":20,"score":77.38,"dimensions":{"reach":77.38,"momentum":77.38},"metrics":{"forks":1349,"stars":8749,"activityScore":99.79}}],"publication":{"scoringVersion":"github-momentum-v1+stars","observedAt":"2026-10-10T02:35:31.561Z","calculatedAt":"2026-10-10T09:57:37.585Z","publishedAt":"2026-10-10T09:57:38.967Z"}}],"editions":[]},{"id":"skills","slug":"skills","primary":"momentum","dimensions":[{"id":"momentum","label":"Dynamique entre sources"},{"id":"reach","label":"Portée des installations Skills.sh"}],"metrics":[{"key":"installs","label":"Installations cumulées","unit":"count","source":"Skills.sh","url":"https://www.skills.sh/docs/api"},{"key":"installs7d","label":"Installations gagnées (7 jours)","unit":"count","source":"Calcul SANDBOX depuis Skills.sh","url":"https://www.skills.sh/docs/api"},{"key":"stars","label":"Étoiles GitHub","unit":"count","source":"GitHub REST API","url":"https://docs.github.com/en/rest/repos/repos#get-a-repository"},{"key":"stars7d","label":"Étoiles gagnées (7 jours)","unit":"count","source":"Calcul SANDBOX depuis GitHub","url":"https://docs.github.com/en/rest/repos/repos#get-a-repository"},{"key":"forks","label":"Forks GitHub","unit":"count","source":"GitHub REST API","url":"https://docs.github.com/en/rest/repos/repos#get-a-repository"},{"key":"freshness","label":"Fraîcheur du dépôt","unit":"score","source":"Calcul SANDBOX depuis GitHub","url":"https://docs.github.com/en/rest/repos/repos#get-a-repository"}],"snapshots":[{"chart":"skills","week":"2026-10-10","publishedAt":"2026-10-10T10:00:31.746Z","retrievedAt":"2026-10-10T10:00:18.421Z","provenance":"auto","entries":[{"entity":"skill-mattpocock-skills-resolving-merge-conflicts-6366401d","rank":1,"score":100,"dimensions":{"reach":100,"momentum":100},"metrics":{"forks":23744,"stars":283469,"installs":550945,"freshness":98.20413008890613}},{"entity":"skill-microsoft-azure-skills-azure-rbac-e2b91e75","rank":2,"score":99.91,"dimensions":{"reach":99.91,"momentum":99.91},"metrics":{"forks":254,"stars":1555,"installs":396677,"freshness":98.8953060074043}},{"entity":"skill-microsoft-azure-skills-azure-cost-e8751621","rank":3,"score":99.83,"dimensions":{"reach":99.83,"momentum":99.83},"metrics":{"forks":254,"stars":1555,"installs":362649,"freshness":98.8953060074043}},{"entity":"skill-microsoft-azure-skills-azure-hosted-copilot-sdk-2cb4c2ac","rank":4,"score":99.74,"dimensions":{"reach":99.74,"momentum":99.74},"metrics":{"forks":254,"stars":1555,"installs":355277,"freshness":98.8953060074043}},{"entity":"skill-mattpocock-skills-to-prd-a1ff2082","rank":5,"score":99.65,"dimensions":{"reach":99.65,"momentum":99.65},"metrics":{"forks":23744,"stars":283469,"installs":351572,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-to-issues-00543cd7","rank":6,"score":99.56,"dimensions":{"reach":99.56,"momentum":99.56},"metrics":{"forks":23744,"stars":283469,"installs":339097,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-writing-great-skills-dbb203e0","rank":7,"score":99.48,"dimensions":{"reach":99.48,"momentum":99.48},"metrics":{"forks":23744,"stars":283469,"installs":306532,"freshness":98.20413008890613}},{"entity":"skill-scrapegraphai-just-scrape-just-scrape-59d9ef84","rank":8,"score":99.39,"dimensions":{"reach":99.39,"momentum":99.39},"metrics":{"forks":14,"stars":69,"installs":244991,"freshness":54.59632005854692}},{"entity":"skill-mattpocock-skills-diagnose-74203c86","rank":9,"score":99.3,"dimensions":{"reach":99.3,"momentum":99.3},"metrics":{"forks":23744,"stars":283469,"installs":227593,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-write-a-skill-aea77044","rank":10,"score":99.22,"dimensions":{"reach":99.22,"momentum":99.22},"metrics":{"forks":23744,"stars":283469,"installs":219742,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-zoom-out-2a509ca1","rank":11,"score":99.13,"dimensions":{"reach":99.13,"momentum":99.13},"metrics":{"forks":23744,"stars":283469,"installs":219142,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-caveman-09c5ab10","rank":12,"score":99.04,"dimensions":{"reach":99.04,"momentum":99.04},"metrics":{"forks":23744,"stars":283469,"installs":214554,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-design-an-interface-c7e98cca","rank":13,"score":98.95,"dimensions":{"reach":98.95,"momentum":98.95},"metrics":{"forks":23744,"stars":283469,"installs":208821,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-request-refactor-plan-693928dd","rank":14,"score":98.87,"dimensions":{"reach":98.87,"momentum":98.87},"metrics":{"forks":23744,"stars":283469,"installs":202423,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-qa-2844017a","rank":15,"score":98.78,"dimensions":{"reach":98.78,"momentum":98.78},"metrics":{"forks":23744,"stars":283469,"installs":198667,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-ubiquitous-language-2a562f12","rank":16,"score":98.69,"dimensions":{"reach":98.69,"momentum":98.69},"metrics":{"forks":23744,"stars":283469,"installs":197194,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-obsidian-vault-d4c5f408","rank":17,"score":98.61,"dimensions":{"reach":98.61,"momentum":98.61},"metrics":{"forks":23744,"stars":283469,"installs":195503,"freshness":98.20413008890613}},{"entity":"skill-mattpocock-skills-edit-article-7b51c229","rank":18,"score":98.52,"dimensions":{"reach":98.52,"momentum":98.52},"metrics":{"forks":23744,"stars":283469,"installs":191771,"freshness":98.20413008890613}},{"entity":"skill-microsoft-azure-skills-azure-cost-optimization-3845cbb2","rank":19,"score":98.43,"dimensions":{"reach":98.43,"momentum":98.43},"metrics":{"forks":254,"stars":1555,"installs":155407,"freshness":98.8953060074043}},{"entity":"skill-heygen-com-hyperframes-hyperframes-media-831e7348","rank":20,"score":98.34,"dimensions":{"reach":98.34,"momentum":98.34},"metrics":{"forks":5361,"stars":60041,"installs":141731,"freshness":100}}],"publication":{"scoringVersion":"skills-installs-v1","observedAt":"2026-10-10T10:00:18.421Z","calculatedAt":"2026-10-10T10:00:30.332Z","publishedAt":"2026-10-10T10:00:31.746Z"}}],"editions":[]},{"id":"models","slug":"ai-models","primary":"quality","dimensions":[{"id":"agents","label":"Agents"},{"id":"coding","label":"Code"},{"id":"multimodal","label":"Vision"},{"id":"quality","label":"Qualité générale"},{"id":"research","label":"Recherche"},{"id":"value","label":"Valeur"}],"metrics":[{"key":"arenaQuality","label":"Elo Arena (texte)","unit":"score","source":"Arena","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"},{"key":"artificialAnalysisIntelligence","label":"Artificial Analysis Intelligence Index","unit":"score","source":"Artificial Analysis via OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/benchmarks/get-benchmarks"},{"key":"arenaCoding","label":"Elo Arena (WebDev)","unit":"score","source":"Arena WebDev","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"},{"key":"arenaAgent","label":"Elo Arena (agents)","unit":"score","source":"Arena Agent","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"},{"key":"arenaVision","label":"Elo Arena (vision)","unit":"score","source":"Arena Vision","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"},{"key":"arenaSearch","label":"Elo Arena (search)","unit":"score","source":"Arena Search","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset"},{"key":"downloads30d","label":"Téléchargements Hugging Face (30 jours)","unit":"count","source":"Hugging Face Hub","url":"https://huggingface.co/docs/hub/api"},{"key":"contextLength","label":"Fenêtre de contexte (tokens)","unit":"tokens","source":"OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/models/get-models"},{"key":"promptPricePerMillion","label":"Prix entrée / million de tokens","unit":"usd","source":"OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/models/get-models"},{"key":"completionPricePerMillion","label":"Prix sortie / million de tokens","unit":"usd","source":"OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/models/get-models"},{"key":"openRouterWeeklyRank","label":"Rang d’usage hebdomadaire OpenRouter","unit":"count","source":"OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/models/get-models"},{"key":"openRouterThroughputRank","label":"Rang de débit OpenRouter","unit":"count","source":"OpenRouter","url":"https://openrouter.ai/docs/api/api-reference/models/get-models"}],"snapshots":[{"chart":"models","week":"2026-W41","publishedAt":"2026-10-10T09:01:36.264Z","retrievedAt":"2026-10-10T09:01:21.376Z","provenance":"auto","entries":[{"entity":"gemini4argonhigh-arena-google-gemini-4-argon-high","rank":1,"score":100,"dimensions":{"value":100,"coding":90.91,"quality":100},"metrics":{"arenaCoding":1677.8058895588372,"arenaQuality":1525.3956435571818}},{"entity":"claudeopus55high-arena-anthropic-claude-opus-5-5-high","rank":2,"score":98.99,"dimensions":{"value":98.99,"quality":98.99,"multimodal":84.85},"metrics":{"arenaVision":1292.6884122295935,"arenaQuality":1507.0758687407206}},{"entity":"claudeopus46high","rank":3,"score":97.98,"dimensions":{"value":97.98,"coding":56.06,"quality":97.98,"multimodal":96.97},"metrics":{"arenaCoding":1546.5481275181767,"arenaVision":1299.2321456198133,"arenaQuality":1504.355209789498}},{"entity":"claudefable5high-arena-anthropic-claude-fable-5-high","rank":4,"score":96.97,"dimensions":{"value":96.97,"coding":78.79,"quality":96.97,"multimodal":100},"metrics":{"arenaCoding":1624.7356714393125,"arenaVision":1307.9658540715964,"arenaQuality":1503.693577172815}},{"entity":"claudeopus47high","rank":5,"score":95.96,"dimensions":{"value":95.96,"coding":62.12,"quality":95.96,"multimodal":93.94},"metrics":{"arenaCoding":1557.4658846362593,"arenaVision":1297.7610703084888,"arenaQuality":1501.0258966973688}},{"entity":"claudefable51max-arena-anthropic-claude-fable-5-1-max","rank":6,"score":94.95,"dimensions":{"value":94.95,"coding":95.45,"quality":94.95,"multimodal":75.76},"metrics":{"arenaCoding":1743.7093865265408,"arenaVision":1286.4074118221424,"arenaQuality":1500.7290291564887}},{"entity":"claudeopus46","rank":7,"score":93.94,"dimensions":{"value":93.94,"coding":50,"quality":93.94,"multimodal":90.91},"metrics":{"arenaCoding":1537.2248745549678,"arenaVision":1294.4596610653605,"arenaQuality":1497.7335923252954}},{"entity":"gemini38flashhigh-arena-google-gemini-3-8-flash-high","rank":8,"score":92.93,"dimensions":{"value":92.93,"coding":66.67,"quality":92.93,"multimodal":80.3},"metrics":{"arenaCoding":1583.462059363355,"arenaVision":1289.1291293859153,"arenaQuality":1496.5039548814862}},{"entity":"musespark13max-arena-meta-muse-spark-1-3-max","rank":9,"score":91.92,"dimensions":{"value":91.92,"coding":86.36,"quality":91.92,"multimodal":83.33},"metrics":{"arenaCoding":1656.8554034202887,"arenaVision":1289.716856872445,"arenaQuality":1494.1854645692001}},{"entity":"claudeopus47","rank":10,"score":90.91,"dimensions":{"value":90.91,"coding":59.09,"quality":90.91,"research":85.71,"multimodal":95.45},"metrics":{"arenaCoding":1555.173246506906,"arenaSearch":1233.2545271621282,"arenaVision":1298.2195689550822,"arenaQuality":1493.9734271816128}},{"entity":"musespark12xhigh-arena-meta-muse-spark-1-2-xhigh","rank":11,"score":89.9,"dimensions":{"value":89.9,"coding":46.97,"quality":89.9,"multimodal":86.36},"metrics":{"arenaCoding":1532.5700125115925,"arenaVision":1292.890549493764,"arenaQuality":1492.0384395808942}},{"entity":"musespark11-arena-meta-muse-spark-1-1","rank":12,"score":88.89,"dimensions":{"value":88.89,"coding":54.55,"quality":88.89,"multimodal":60.61},"metrics":{"arenaCoding":1542.1189310438444,"arenaVision":1279.9549965259757,"arenaQuality":1490.709929717255}},{"entity":"claudeopus5high-arena-anthropic-claude-opus-5-high","rank":13,"score":87.88,"dimensions":{"value":87.88,"coding":87.88,"quality":87.88,"multimodal":78.79},"metrics":{"arenaCoding":1656.8736934757285,"arenaVision":1287.4166281489183,"arenaQuality":1490.128697977279}},{"entity":"claudeopus5max-arena-anthropic-claude-opus-5-max","rank":14,"score":86.87,"dimensions":{"value":86.87,"coding":93.94,"quality":86.87},"metrics":{"arenaCoding":1691.3111706293098,"arenaQuality":1489.409519068171}},{"entity":"musespark","rank":15,"score":85.86,"dimensions":{"value":85.86,"quality":85.86,"multimodal":87.88},"metrics":{"arenaVision":1293.8704719961554,"arenaQuality":1488.883450740479}},{"entity":"kimik3max-arena-moonshot-kimi-k3-max","rank":16,"score":84.85,"dimensions":{"value":84.85,"coding":84.85,"quality":84.85},"metrics":{"arenaCoding":1654.354241193575,"arenaQuality":1488.0769526460258}},{"entity":"gemini31propreview-arena-google-gemini-3-1-pro-preview","rank":17,"score":83.84,"dimensions":{"value":83.84,"coding":24.24,"quality":83.84,"multimodal":59.09},"metrics":{"arenaCoding":1446.7086676853123,"arenaVision":1279.7787313231265,"arenaQuality":1487.3024091972857}},{"entity":"gemini37flashhigh-arena-google-gemini-3-7-flash-high","rank":18,"score":82.83,"dimensions":{"value":82.83,"coding":68.18,"quality":82.83,"multimodal":92.42},"metrics":{"arenaCoding":1591.5464842105637,"arenaVision":1296.8746588285421,"arenaQuality":1486.4994828635868}},{"entity":"gemini3pro","rank":19,"score":81.82,"dimensions":{"value":81.82,"coding":22.73,"quality":81.82,"multimodal":81.82},"metrics":{"arenaCoding":1439.5236906232656,"arenaVision":1289.3237216009584,"arenaQuality":1485.492554247684}},{"entity":"gpt56solxhigh-arena-openai-gpt-5-6-sol-xhigh","rank":20,"score":80.81,"dimensions":{"value":80.81,"quality":80.81,"research":100,"multimodal":77.27},"metrics":{"arenaSearch":1257.2603500980206,"arenaVision":1286.9619006302116,"arenaQuality":1485.066674207078}}],"publication":{"scoringVersion":"models-cross-source-v1","observedAt":"2026-10-10T09:01:21.376Z","calculatedAt":"2026-10-10T09:01:34.884Z","publishedAt":"2026-10-10T09:01:36.264Z"}},{"chart":"models","week":"2026-10-11","publishedAt":"2026-10-11T03:33:01.232Z","retrievedAt":"2026-10-11T02:39:33.061Z","provenance":"auto","entries":[{"entity":"claudeopus55maxdefaultfallback","rank":1,"score":100,"dimensions":{"value":100,"quality":100},"metrics":{"artificialAnalysisIntelligence":57.6}},{"entity":"gemini4argonhigh-arena-google-gemini-4-argon-high","rank":2,"score":100,"dimensions":{"value":100,"coding":91.55,"quality":100},"metrics":{"arenaCoding":1677.8058895588372,"arenaQuality":1525.3956435571818}},{"entity":"claudesonnet55maxdefaultfallback","rank":3,"score":99.71,"dimensions":{"value":99.71,"quality":99.71},"metrics":{"artificialAnalysisIntelligence":56}},{"entity":"claudefable51maxdefaultfallback","rank":4,"score":99.28,"dimensions":{"value":99.28,"agents":100,"coding":100,"quality":99.28},"metrics":{"artificialAnalysisCoding":81.6,"artificialAnalysisAgentic":57.9,"artificialAnalysisIntelligence":53.4}},{"entity":"qwen38max-artificial-analysis-qwen-qwen3-8-max-20260803","rank":5,"score":99.28,"dimensions":{"value":99.28,"agents":87.88,"coding":81.69,"quality":99.28},"metrics":{"artificialAnalysisCoding":68.9,"artificialAnalysisAgentic":49.9,"artificialAnalysisIntelligence":53.4}},{"entity":"claudeopus55high-arena-anthropic-claude-opus-5-5-high","rank":6,"score":98.99,"dimensions":{"value":98.99,"quality":98.99,"multimodal":85.92},"metrics":{"arenaVision":1292.6884122295935,"arenaQuality":1507.0758687407206}},{"entity":"gpt6astramax-artificial-analysis-openai-gpt-6-astra-20260903","rank":7,"score":98.84,"dimensions":{"value":98.84,"agents":93.94,"coding":97.89,"quality":98.84},"metrics":{"artificialAnalysisCoding":76.9,"artificialAnalysisAgentic":51,"artificialAnalysisIntelligence":52.7}},{"entity":"gpt61solmax-artificial-analysis-openai-gpt-6-1-sol-20260929","rank":8,"score":98.55,"dimensions":{"value":98.55,"quality":98.55},"metrics":{"artificialAnalysisIntelligence":51.8}},{"entity":"claudeopus5max-artificial-analysis-anthropic-claude-opus-5-20260723","rank":9,"score":98.26,"dimensions":{"value":98.26,"agents":98.99,"coding":99.3,"quality":98.26},"metrics":{"artificialAnalysisCoding":78,"artificialAnalysisAgentic":56.5,"artificialAnalysisIntelligence":50.8}},{"entity":"claudeopus46high","rank":10,"score":97.98,"dimensions":{"value":97.98,"coding":57.75,"quality":97.98,"multimodal":97.18},"metrics":{"arenaCoding":1546.5481275181767,"arenaVision":1299.2321456198133,"arenaQuality":1504.355209789498}},{"entity":"claudefable5maxopus48fallback","rank":11,"score":97.97,"dimensions":{"value":97.97,"agents":91.92,"coding":95.77,"quality":97.97},"metrics":{"artificialAnalysisCoding":76.5,"artificialAnalysisAgentic":50.7,"artificialAnalysisIntelligence":49.6}},{"entity":"musespark13max-artificial-analysis-meta-muse-spark-1-3-20260902","rank":12,"score":97.68,"dimensions":{"value":97.68,"agents":96.97,"coding":92.96,"quality":97.68},"metrics":{"artificialAnalysisCoding":75.8,"artificialAnalysisAgentic":55.5,"artificialAnalysisIntelligence":48.1}},{"entity":"gpt6solmax-artificial-analysis-openai-gpt-6-sol-20260922","rank":13,"score":97.39,"dimensions":{"value":97.39,"quality":97.39},"metrics":{"artificialAnalysisIntelligence":47.6}},{"entity":"gpt56solmax","rank":14,"score":97.1,"dimensions":{"value":97.1,"agents":90.91,"coding":98.59,"quality":97.1},"metrics":{"artificialAnalysisCoding":77.4,"artificialAnalysisAgentic":50.2,"artificialAnalysisIntelligence":47}},{"entity":"claudefable5high-arena-anthropic-claude-fable-5-high","rank":15,"score":96.97,"dimensions":{"value":96.97,"coding":78.87,"quality":96.97,"multimodal":100},"metrics":{"arenaCoding":1624.7356714393125,"arenaVision":1307.9658540715964,"arenaQuality":1503.693577172815}},{"entity":"grok47xhigh-artificial-analysis-x-ai-grok-4-7-20260916","rank":16,"score":96.81,"dimensions":{"value":96.81,"quality":96.81},"metrics":{"artificialAnalysisIntelligence":46.4}},{"entity":"mimov26pro-artificial-analysis-xiaomi-mimo-v2-6-pro-20260921","rank":17,"score":96.52,"dimensions":{"value":96.52,"quality":96.52},"metrics":{"artificialAnalysisIntelligence":46.3}},{"entity":"qwen38max0902","rank":18,"score":96.23,"dimensions":{"value":96.23,"agents":97.98,"coding":92.08,"quality":96.23},"metrics":{"arenaCoding":1673.5014803771517,"artificialAnalysisCoding":76.2,"artificialAnalysisAgentic":56,"artificialAnalysisIntelligence":45.4}},{"entity":"claudeopus47high","rank":19,"score":95.96,"dimensions":{"value":95.96,"coding":63.38,"quality":95.96,"multimodal":94.37},"metrics":{"arenaCoding":1557.4658846362593,"arenaVision":1297.7610703084888,"arenaQuality":1501.0258966973688}},{"entity":"glm53max-artificial-analysis-z-ai-glm-5-3-20260816","rank":20,"score":95.94,"dimensions":{"value":95.94,"agents":95.96,"coding":91.55,"quality":95.94},"metrics":{"artificialAnalysisCoding":74.8,"artificialAnalysisAgentic":53.1,"artificialAnalysisIntelligence":44.8}}],"publication":{"scoringVersion":"models-cross-source-v1","observedAt":"2026-10-11T02:39:33.061Z","calculatedAt":"2026-10-11T03:33:00.978Z","publishedAt":"2026-10-11T03:33:01.232Z"}},{"chart":"models","week":"2026-10-12","publishedAt":"2026-10-12T03:33:00.636Z","retrievedAt":"2026-10-12T02:39:32.496Z","provenance":"auto","entries":[{"entity":"claudeopus55maxdefaultfallback","rank":1,"score":100,"dimensions":{"value":100,"quality":100},"metrics":{"artificialAnalysisIntelligence":57.6}},{"entity":"gemini4argonhigh-arena-google-gemini-4-argon-high","rank":2,"score":100,"dimensions":{"value":100,"coding":91.55,"quality":100},"metrics":{"arenaCoding":1677.8058895588372,"arenaQuality":1525.3956435571818}},{"entity":"claudesonnet55maxdefaultfallback","rank":3,"score":99.71,"dimensions":{"value":99.71,"quality":99.71},"metrics":{"artificialAnalysisIntelligence":56}},{"entity":"claudefable51maxdefaultfallback","rank":4,"score":99.28,"dimensions":{"value":99.28,"agents":100,"coding":100,"quality":99.28},"metrics":{"artificialAnalysisCoding":81.6,"artificialAnalysisAgentic":57.9,"artificialAnalysisIntelligence":53.4}},{"entity":"qwen38max-artificial-analysis-qwen-qwen3-8-max-20260803","rank":5,"score":99.28,"dimensions":{"value":99.28,"agents":87.88,"coding":81.69,"quality":99.28},"metrics":{"artificialAnalysisCoding":68.9,"artificialAnalysisAgentic":49.9,"artificialAnalysisIntelligence":53.4}},{"entity":"claudeopus55high-arena-anthropic-claude-opus-5-5-high","rank":6,"score":98.99,"dimensions":{"value":98.99,"quality":98.99,"multimodal":85.92},"metrics":{"arenaVision":1292.6884122295935,"arenaQuality":1507.0758687407206}},{"entity":"gpt6astramax-artificial-analysis-openai-gpt-6-astra-20260903","rank":7,"score":98.84,"dimensions":{"value":98.84,"agents":93.94,"coding":97.89,"quality":98.84},"metrics":{"artificialAnalysisCoding":76.9,"artificialAnalysisAgentic":51,"artificialAnalysisIntelligence":52.7}},{"entity":"gpt61solmax-artificial-analysis-openai-gpt-6-1-sol-20260929","rank":8,"score":98.55,"dimensions":{"value":98.55,"quality":98.55},"metrics":{"artificialAnalysisIntelligence":51.8}},{"entity":"claudeopus5max-artificial-analysis-anthropic-claude-opus-5-20260723","rank":9,"score":98.26,"dimensions":{"value":98.26,"agents":98.99,"coding":99.3,"quality":98.26},"metrics":{"artificialAnalysisCoding":78,"artificialAnalysisAgentic":56.5,"artificialAnalysisIntelligence":50.8}},{"entity":"claudeopus46high","rank":10,"score":97.98,"dimensions":{"value":97.98,"coding":57.75,"quality":97.98,"multimodal":97.18},"metrics":{"arenaCoding":1546.5481275181767,"arenaVision":1299.2321456198133,"arenaQuality":1504.355209789498}},{"entity":"claudefable5maxopus48fallback","rank":11,"score":97.97,"dimensions":{"value":97.97,"agents":91.92,"coding":95.77,"quality":97.97},"metrics":{"artificialAnalysisCoding":76.5,"artificialAnalysisAgentic":50.7,"artificialAnalysisIntelligence":49.6}},{"entity":"musespark13max-artificial-analysis-meta-muse-spark-1-3-20260902","rank":12,"score":97.68,"dimensions":{"value":97.68,"agents":96.97,"coding":92.96,"quality":97.68},"metrics":{"artificialAnalysisCoding":75.8,"artificialAnalysisAgentic":55.5,"artificialAnalysisIntelligence":48.1}},{"entity":"gpt6solmax-artificial-analysis-openai-gpt-6-sol-20260922","rank":13,"score":97.39,"dimensions":{"value":97.39,"quality":97.39},"metrics":{"artificialAnalysisIntelligence":47.6}},{"entity":"gpt56solmax","rank":14,"score":97.1,"dimensions":{"value":97.1,"agents":90.91,"coding":98.59,"quality":97.1},"metrics":{"artificialAnalysisCoding":77.4,"artificialAnalysisAgentic":50.2,"artificialAnalysisIntelligence":47}},{"entity":"claudefable5high-arena-anthropic-claude-fable-5-high","rank":15,"score":96.97,"dimensions":{"value":96.97,"coding":78.87,"quality":96.97,"multimodal":100},"metrics":{"arenaCoding":1624.7356714393125,"arenaVision":1307.9658540715964,"arenaQuality":1503.693577172815}},{"entity":"grok47xhigh-artificial-analysis-x-ai-grok-4-7-20260916","rank":16,"score":96.81,"dimensions":{"value":96.81,"quality":96.81},"metrics":{"artificialAnalysisIntelligence":46.4}},{"entity":"mimov26pro-artificial-analysis-xiaomi-mimo-v2-6-pro-20260921","rank":17,"score":96.52,"dimensions":{"value":96.52,"quality":96.52},"metrics":{"artificialAnalysisIntelligence":46.3}},{"entity":"qwen38max0902","rank":18,"score":96.23,"dimensions":{"value":96.23,"agents":97.98,"coding":92.08,"quality":96.23},"metrics":{"arenaCoding":1673.5014803771517,"artificialAnalysisCoding":76.2,"artificialAnalysisAgentic":56,"artificialAnalysisIntelligence":45.4}},{"entity":"claudeopus47high","rank":19,"score":95.96,"dimensions":{"value":95.96,"coding":63.38,"quality":95.96,"multimodal":94.37},"metrics":{"arenaCoding":1557.4658846362593,"arenaVision":1297.7610703084888,"arenaQuality":1501.0258966973688}},{"entity":"glm53max-artificial-analysis-z-ai-glm-5-3-20260816","rank":20,"score":95.94,"dimensions":{"value":95.94,"agents":95.96,"coding":91.55,"quality":95.94},"metrics":{"artificialAnalysisCoding":74.8,"artificialAnalysisAgentic":53.1,"artificialAnalysisIntelligence":44.8}}],"publication":{"scoringVersion":"models-cross-source-v1","observedAt":"2026-10-12T02:39:32.496Z","calculatedAt":"2026-10-12T03:33:00.399Z","publishedAt":"2026-10-12T03:33:00.636Z"}}],"editions":[]},{"id":"rising","slug":"rising","primary":"discovery","dimensions":[{"id":"discovery","label":"Discovery"}],"metrics":[{"key":"stars","label":"Étoiles","unit":"count","source":"GitHub API","url":"https://docs.github.com/en/graphql/reference/repos"},{"key":"forks","label":"Forks","unit":"count","source":"GitHub API","url":"https://docs.github.com/en/graphql/reference/repos"},{"key":"contributors","label":"Contributeurs","unit":"count","source":"GitHub API","url":"https://docs.github.com/en/graphql/reference/repos"},{"key":"stars7d","label":"Étoiles gagnées (7 jours)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/rising/methodology"},{"key":"growth","label":"Croissance (%)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/rising/methodology"},{"key":"forks7d","label":"Forks gagnés (7 jours)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/rising/methodology"},{"key":"contributorsDelta","label":"Évolution des contributeurs","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/rising/methodology"},{"key":"activityScore","label":"Activité du dépôt","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/rising/methodology"},{"key":"acceleration","label":"Accélération (%)","unit":"count","source":"Calcul SANDBOX sur les relevés GitHub","url":"/charts/rising/methodology"}],"snapshots":[],"editions":[]}],"takes":[],"episodes":[],"newsletterUrl":null,"progress":{"databaseReady":true,"trackedRepositories":1328,"snapshotsToday":1600,"distinctSnapshotDays":6,"requiredHistoryDays":7,"lastSuccessfulCollectAt":"2026-10-15 09:45:45.175+00","earliestPossibleEditionWeek":"2026-10-17"},"badgeSlugs":[]},"initialWeek":"2026-W41","initialSelection":{"period":"week"}}] 8:null d:[["$","title","0",{"children":"Ranking archives · 2026-W41 · SANDBOX CHARTS · Sandbox"}],["$","meta","1",{"name":"description","content":"Les positions et les mouvements de l'édition 2026-W41."}],["$","meta","2",{"name":"application-name","content":"Sandbox"}],["$","meta","3",{"name":"robots","content":"index, follow, max-video-preview:-1, max-image-preview:large"}],["$","meta","4",{"name":"googlebot","content":"index, follow, max-video-preview:-1, max-image-preview:large"}],["$","link","5",{"rel":"canonical","href":"https://www.sandboxpodcast.fr/en/charts/history/2026-W41"}],["$","meta","6",{"property":"og:title","content":"Ranking archives · 2026-W41 · SANDBOX CHARTS"}],["$","meta","7",{"property":"og:description","content":"Les positions et les mouvements de l'édition 2026-W41."}],["$","meta","8",{"property":"og:url","content":"https://www.sandboxpodcast.fr/en/charts/history/2026-W41"}],["$","meta","9",{"property":"og:site_name","content":"Sandbox"}],["$","meta","10",{"property":"og:locale","content":"en"}],["$","meta","11",{"property":"og:image","content":"https://www.sandboxpodcast.fr/api/social-image"}],["$","meta","12",{"property":"og:image:width","content":"1200"}],["$","meta","13",{"property":"og:image:height","content":"630"}],["$","meta","14",{"property":"og:image:alt","content":"Sandbox : Talk Show et classements tech"}],["$","meta","15",{"property":"og:type","content":"website"}],["$","meta","16",{"name":"twitter:card","content":"summary_large_image"}],["$","meta","17",{"name":"twitter:title","content":"Ranking archives · 2026-W41 · SANDBOX CHARTS"}],["$","meta","18",{"name":"twitter:description","content":"Les positions et les mouvements de l'édition 2026-W41."}],["$","meta","19",{"name":"twitter:image","content":"https://www.sandboxpodcast.fr/api/social-image"}],["$","meta","20",{"name":"twitter:image:alt","content":"Sandbox : Talk Show et classements tech"}],["$","link","21",{"rel":"icon","href":"/icon.svg?icon.3h3k7k9gyk_88.svg","sizes":"any","type":"image/svg+xml"}],["$","link","22",{"rel":"apple-touch-icon","href":"/apple-icon?7296b06b86efdbda","alt":"$undefined","type":"image/png","sizes":"180x180"}],["$","$L21","23",{}]]