Pass IndexThe State of AISign in

Proofbench — Epoch AI

Proofbench — Epoch AI is run by Epoch AI. It has ranked 64 models, of which the catalogue holds 62, scoring from 0 to 100.

measured by Epoch AI · the board itself

PlaceModelMetricScore
1stof 64Claude Fable 5.1Accuracy (max)100
2ndof 64Claude Opus 5Accuracy (max)99
3rdof 64GPT 6 AstraAccuracy99
4thof 64Claude Fable 5Accuracy (max)95
5thof 64Kimi K3Accuracy87
6thof 64GPT-5.6 SolAccuracy (max)83
7thof 64Claude Sonnet 5Accuracy (max)77
8thof 64GPT-5.6 TerraAccuracy (xhigh)74
9thof 64Claude Opus 4.8Accuracy (max)69
10thof 64GPT-5.6 LunaAccuracy (max)60
11thof 64Gemini 3.7 FlashAccuracy58
12thof 64Qwen 3.8 MaxAccuracy58
13thof 64GPT-5.4Accuracy (xhigh)56
14thof 64DeepSeek V4 Flash (0731)Accuracy56
15thof 64Claude Opus 4.7Accuracy (max)54
16thof 64Grok 4.6Accuracy51
17thof 64GPT-5.5Accuracy (xhigh)50
18thof 64Claude Opus 4.6Accuracy (max)50
19thof 64DeepSeek V4 Pro 0813Accuracy50
20thof 64GLM-5.3Accuracy (max)49
21stof 64Gemini 3.8 FlashAccuracy48
22ndof 64Claude Sonnet 4.6Accuracy (max)45
23rdof 64Muse Spark 1.2Accuracy43
24thof 64Muse Spark 1.1Accuracy39
25thof 64Claude Opus 4.5Accuracy36
26thof 64Gemini 3.6 FlashAccuracy36
27thof 64GLM 5.2Accuracy (max)35
28thof 64Grok 4.5Accuracy (high)31
29thof 64Gemini 3.5 FlashAccuracy (high)31
30thof 64Qwen3.7 MaxAccuracy (max)26
31stof 64Gemini 3.1 Pro PreviewAccuracy26
32ndof 64GLM 5.1Accuracy22.222
33rdof 64MiMo-V2.5-ProAccuracy22
34thof 64GPT-5.4 miniAccuracy (xhigh)21
35thof 64GLM 5.3 FlashAccuracy (max)21
36thof 64gemini-3-pro-previewAccuracy20
37thof 64Claude Sonnet 4.5Accuracy19
38thof 64MiniMax M3Accuracy18
39thof 64GPT-5Accuracy (high)18
41stof 64Kimi K2.6Accuracy16
42ndof 64MiMo-V2.5Accuracy16
43rdof 64DeepSeek V4 ProAccuracy (max)16
44thof 64Qwen3.8 27BAccuracy16
45thof 64Gemini 3 Flash PreviewAccuracy15
46thof 64GPT-5.2Accuracy (xhigh)15
47thof 64Grok 4.20Accuracy14
48thof 64Gemini 3.5 Flash-LiteAccuracy13
49thof 64GPT-5 nanoAccuracy (high)12
50thof 64Grok 4.3Accuracy (high)11
51stof 64GPT-5 miniAccuracy (high)9
52ndof 64GPT-5.1 Codex MaxAccuracy (max)9
54thof 64DeepSeek v3.2Accuracy8
55thof 64GLM-4.7Accuracy6
56thof 64Inkling SmallAccuracy6
57thof 64GPT-5.4 nanoAccuracy (high)5
58thof 64Grok 4.1 Fast ReasoningAccuracy4
59thof 64MiniMax M2.5Accuracy4
60thof 64MiniMax M2.7Accuracy3
61stof 64Nemotron 3 UltraAccuracy2
62ndof 64Laguna-XS.2Accuracy0
63rdof 64Laguna M.1Accuracy0
64thof 64InklingAccuracy0

Read from the board on 2026-09-16