AI model rankings
Discover tools and use the data to compare them.
Dated, sourced editions
MODELS TOP 20
Compare models by the capability you care about.
Understand this edition's data
This edition ranks models that have an admissible Arena evaluation, an admissible Artificial Analysis index, or both. A missing evaluation is not estimated and is not scored as 100. Two models at 100 may each be the best result of a different test. This score is not the corroborated v3 profile.
- Latest included observation
- Frozen calculation
- Publication
Calculation version : models-quality-observed
20 entries shown in MODELS TOP 20.
01-
Claude Opus 5.5 (Max, Default Fallback)
Claude Opus 5.5 (Max, Default Fallback)
100,0/100SANDBOX SCOREIntroduction
Models
Claude Opus 5.5 (Max, Default Fallback)
Data
SANDBOX score: 100,0 out of 100.
Measurements, score components and sources
Score components
- Observed evaluations
- 100,0/100
02-
Claude Sonnet 5.5 (Max, Default Fallback)
Claude Sonnet 5.5 (Max, Default Fallback)
99,7/100SANDBOX SCOREIntroduction
Models
Claude Sonnet 5.5 (Max, Default Fallback)
Data
SANDBOX score: 99,7 out of 100.
Measurements, score components and sources
Score components
- Observed evaluations
- 99,7/100
03-
Claude Fable 5.1 (Max, Default Fallback)
Claude Fable 5.1 (Max, Default Fallback)
99,2/100SANDBOX SCOREIntroduction
Models
Claude Fable 5.1 (Max, Default Fallback)
Data
SANDBOX score: 99,2 out of 100.
Measurements, score components and sources
Score components
- Observed evaluations
- 99,2/100
04-
Qwen3.8 Max
Qwen3.8 Max
99,2/100SANDBOX SCOREIntroduction
Models
Qwen3.8 Max
Data
SANDBOX score: 99,2 out of 100.
Measurements, score components and sources
Score components
- Observed evaluations
- 99,2/100
05-
Gemini 4 Argon (High)
Gemini 4 Argon (High)
99,1/100SANDBOX SCORE-Introduction
Models
Gemini 4 Argon (High)
Data
SANDBOX score: 99,1 out of 100.
30 DAYS MOMENTUM
History is being collectedW41-11.10-12.10-19.10 00h#51 editions in the Top 20 · best rank #5
Measurements, score components and sources
- Disagreement between families
- 0,8 Calculated from collected measurements
Score components
- Observed evaluations
- 99,1/100
06-
Muse Wonder (Max)
Muse Wonder (Max)
98,9/100SANDBOX SCORE-Introduction
Models
Muse Wonder (Max)
Data
SANDBOX score: 98,9 out of 100.
30 DAYS MOMENTUM
History is being collectedW41-11.10-12.10-19.10 00h#61 editions in the Top 20 · best rank #6
Measurements, score components and sources
Score components
- Observed evaluations
- 98,9/100
07-
GPT-6 Astra (Max)
GPT-6 Astra (Max)
98,8/100SANDBOX SCOREIntroduction
Models
GPT-6 Astra (Max)
Data
SANDBOX score: 98,8 out of 100.
Measurements, score components and sources
Score components
- Observed evaluations
- 98,8/100
08-
GPT-6.1 Sol (Max)
GPT-6.1 Sol (Max)
98,2/100SANDBOX SCOREIntroduction
Models
GPT-6.1 Sol (Max)
Data
SANDBOX score: 98,2 out of 100.
Measurements, score components and sources
Score components
- Observed evaluations
- 98,2/100
09-
Claude Opus 5.5 (High)
Claude Opus 5.5 (High)
97,8/100SANDBOX SCORE-Introduction
Models
Claude Opus 5.5 (High)
Data
SANDBOX score: 97,8 out of 100.
30 DAYS MOMENTUM
History is being collectedW41-11.10-12.10-19.10 00h#91 editions in the Top 20 · best rank #9
Measurements, score components and sources
Score components
- Observed evaluations
- 97,8/100
10-
Claude Fable 5 (Max, Opus 4.8 Fallback)
Claude Fable 5 (Max, Opus 4.8 Fallback)
97,6/100SANDBOX SCOREIntroduction
Models
Claude Fable 5 (Max, Opus 4.8 Fallback)
Data
SANDBOX score: 97,6 out of 100.
30 DAYS MOMENTUM
W41-11.10#1112.10#1119.10 00h#103 editions in the Top 20 · best rank #10
Measurements, score components and sources
Score components
- Observed evaluations
- 97,6/100
11-
Muse Spark 1.3 (Max)
Muse Spark 1.3 (Max)
97,3/100SANDBOX SCOREIntroduction
Models
Muse Spark 1.3 (Max)
Data
SANDBOX score: 97,3 out of 100.
30 DAYS MOMENTUM
W41-11.10#1212.10#1219.10 00h#113 editions in the Top 20 · best rank #11
Measurements, score components and sources
Score components
- Observed evaluations
- 97,3/100
12-
GPT-6 Sol (Max)
GPT-6 Sol (Max)
97,0/100SANDBOX SCOREIntroduction
Models
GPT-6 Sol (Max)
Data
SANDBOX score: 97,0 out of 100.
30 DAYS MOMENTUM
W41-11.10#1312.10#1319.10 00h#123 editions in the Top 20 · best rank #12
Measurements, score components and sources
Score components
- Observed evaluations
- 97,0/100
13-
GPT-5.6 Sol (Max)
GPT-5.6 Sol (Max)
96,7/100SANDBOX SCOREIntroduction
Models
GPT-5.6 Sol (Max)
Data
SANDBOX score: 96,7 out of 100.
30 DAYS MOMENTUM
W41-11.10#1412.10#1419.10 00h#133 editions in the Top 20 · best rank #13
Measurements, score components and sources
Score components
- Observed evaluations
- 96,7/100
14-
Claude Fable 5 (High)
Claude Fable 5 (High)
96,7/100SANDBOX SCORE-Introduction
Models
Claude Fable 5 (High)
Data
SANDBOX score: 96,7 out of 100.
30 DAYS MOMENTUM
History is being collectedW41-11.10-12.10-19.10 00h#141 editions in the Top 20 · best rank #14
Measurements, score components and sources
Score components
- Observed evaluations
- 96,7/100
15-
Grok 4.7 (Xhigh)
Grok 4.7 (Xhigh)
96,4/100SANDBOX SCOREIntroduction
Models
Grok 4.7 (Xhigh)
Data
SANDBOX score: 96,4 out of 100.
30 DAYS MOMENTUM
W41-11.10#1612.10#1619.10 00h#153 editions in the Top 20 · best rank #15
Measurements, score components and sources
Score components
- Observed evaluations
- 96,4/100
16-
MiMo-V2.6-Pro
MiMo-V2.6-Pro
96,1/100SANDBOX SCOREIntroduction
Models
MiMo-V2.6-Pro
Data
SANDBOX score: 96,1 out of 100.
30 DAYS MOMENTUM
W41-11.10#1712.10#1719.10 00h#163 editions in the Top 20 · best rank #16
Measurements, score components and sources
Score components
- Observed evaluations
- 96,1/100
17-
GLM-5.3 (Max)
GLM-5.3 (Max)
95,8/100SANDBOX SCOREIntroduction
Models
GLM-5.3 (Max)
Data
SANDBOX score: 95,8 out of 100.
30 DAYS MOMENTUM
W41-11.10#2012.10#2019.10 00h#173 editions in the Top 20 · best rank #17
Measurements, score components and sources
Score components
- Observed evaluations
- 95,8/100
18-
Claude Opus 4.6 (High)
Claude Opus 4.6 (High)
95,6/100SANDBOX SCORE-Introduction
Models
Claude Opus 4.6 (High)
Data
SANDBOX score: 95,6 out of 100.
30 DAYS MOMENTUM
History is being collectedW41-11.10-12.10-19.10 00h#181 editions in the Top 20 · best rank #18
Measurements, score components and sources
Score components
- Observed evaluations
- 95,6/100
19-
Claude Haiku 5.5 (Max)
Claude Haiku 5.5 (Max)
94,9/100SANDBOX SCORE-Introduction
Models
Claude Haiku 5.5 (Max)
Data
SANDBOX score: 94,9 out of 100.
30 DAYS MOMENTUM
History is being collectedW41-11.10-12.10-19.10 00h#191 editions in the Top 20 · best rank #19
Measurements, score components and sources
Score components
- Observed evaluations
- 94,9/100
20-
GPT-5.6 Terra (Max)
GPT-5.6 Terra (Max)
94,6/100SANDBOX SCORE-Introduction
Models
GPT-5.6 Terra (Max)
Data
SANDBOX score: 94,6 out of 100.
30 DAYS MOMENTUM
History is being collectedW41-11.10-12.10-19.10 00h#201 editions in the Top 20 · best rank #20
Measurements, score components and sources
Score components
- Observed evaluations
- 94,6/100
