Epoch Capabilities Index

Epoch AI's own composite capability score fitted across many benchmarks. Shown for reference; not part of the BenchLeader Index.

Published by

Epoch AI Benchmarking Hub

Category

Composite

Index weight

Reference only

Models

235

Data as of

9 Sept 2026

CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.

1. 1GPT-6 Astra166.6
2. 2Claude Fable 5.1164.2
3. 3Claude Fable 5163.4
4. 4Claude Opus 5162.6
5. 5GPT-5.5 Pro162.3
6. 6GPT-5.6 Sol161.8
7. 7GPT-5.6 Terra159.2
8. 8GPT-5.5159.1
9. 9GPT-5.4 Pro158.9
10. 10Claude Opus 4.8158.3
11. 11Kimi K3157.6
12. 12Gemini 3.7 Flash157.4
13. 13GPT-5.4156.9
14. 14Qwen3 8156.6
15. 15GPT-5.3 Codex156.6

SearchAll providers01.AIAlibabaAmazonAnthropicBaichuanCerebras SystemsCohereDatabricksDeepSeekGoogleLMSYSMetaMicrosoftMiniMaxMistral AIMoonshot AIMosaicMLNVIDIAOpenAIPrime IntellectSalesforceShanghai AI LabStability AITechnology Innovation InstituteThinking MachinesUnknownxAIZhipu AIOpen weights235 of 235

#	Model	Score ↓	Run ↕	Index ↕	Published ↕
1	GPT-6 AstraOpenAI	166.6	GPT-6 Astra	69.9	–
2	Claude Fable 5.1Anthropic	164.2	Claude Fable 5.1	70.0	–
3	Claude Fable 5Anthropic	163.4	Claude Fable 5	70.4	–
4	Claude Opus 5Anthropic	162.6	Claude Opus 5	70.0	–
5	GPT-5.5 ProOpenAI	162.3	GPT-5.5 Pro	–	–
6	GPT-5.6 SolOpenAI	161.8	GPT-5.6 Sol	65.3	–
7	GPT-5.6 TerraOpenAI	159.2	GPT-5.6 Terra	62.3	–
8	GPT-5.5OpenAI	159.1	GPT-5.5	66.6	–
9	GPT-5.4 ProOpenAI	158.9	GPT-5.4 Pro	64.2	–
10	Claude Opus 4.8Anthropic	158.3	Claude Opus 4.8	64.7	–
11	Kimi K3Moonshot AIopen	157.6	Kimi K3	66.5	–
12	Gemini 3.7 FlashGoogle	157.4	Gemini 3.7 Flash	61.7	–
13	GPT-5.4OpenAI	156.9	GPT-5.4	63.4	–
14	Qwen3 8Alibaba	156.6	–	63.9	–
15	GPT-5.3 CodexOpenAI	156.6	GPT-5.3 Codex	62.4	–
16	Claude Opus 4.7Anthropic	156.3	Claude Opus 4.7	65.5	–
17	Grok 4.6xAI	156.3	Grok 4.6	63.6	–
18	GPT-5.6 LunaOpenAI	156.3	GPT-5.6 Luna	56.6	–
19	Claude Sonnet 5Anthropic	156.2	Claude Sonnet 5	61.2	–
20	Muse Spark 1.2Meta	155.5	Muse Spark 1.2	62.6	–
21	DeepSeek V4 ProDeepSeekopen	155.4	0813	59.7	–
22	Claude Opus 4.6Anthropic	155.3	Claude Opus 4.6	63.1	–
23	GPT-5.2 ProOpenAI	155.3	GPT-5.2 Pro	–	–
24	GLM-5.3Zhipu AIopen	155.3	GLM-5.3	–	–
25	Gemini 3.1 ProGoogle	155	Gemini 3.1 Pro	63.9	–
26	Gemini 3.5 FlashGoogle	154.7	Gemini 3.5 Flash	60.7	–
27	Muse Spark 1.1Meta	154.6	Muse Spark 1.1	64.9	–
28	DeepSeek V4 FlashDeepSeekopen	154.5	0731	59.1	–
29	Gemini 3.6 FlashGoogle	154.3	Gemini 3.6 Flash	60.7	–
30	Grok 4.5xAI	153.9	Grok 4.5	63.4	–
31	Qwen3 7Alibaba	153.7	–	58.9	–
32	GPT-5.2OpenAI	153.5	GPT-5.2	61.5	–
33	Gemini 3 ProGoogle	153.0	Gemini 3 Pro	61.6	–
34	Claude Sonnet 4.6Anthropic	152.3	Claude Sonnet 4.6	60.3	–
35	Muse SparkMeta	152.1	–	65.4	–
36	Grok 4.20xAI	152.0	Grok 4.20	57.9	–
37	GLM-5.2Zhipu AIopen	151.9	GLM-5.2	57.4	–
38	Gemini 3 FlashGoogle	151.8	Gemini 3 Flash	56.6	–
39	GLM-5.3-FlashZhipu AIopen	151.4	–	61.0	–
40	Kimi K2.6Moonshot AIopen	151.0	Kimi K2.6	60.6	–
41	GPT-5 ProOpenAI	150.3	GPT-5 Pro	60.7	–
42	TInkling-SmallThinking Machinesopen	150.2	–	55.0	–
43	Claude Opus 4.5Anthropic	150.1	Claude Opus 4.5	58.0	–
44	Kimi K2.7 CodeMoonshot AIopen	150.1	Kimi K2.7 Code	56.2	–
45	GPT-5OpenAI	150	GPT-5	59.5	–
46	GLM-5.1Zhipu AIopen	149.7	GLM-5.1	60.1	–
47	GPT-5.1OpenAI	149.7	GPT-5.1	59.4	–
48	Qwen3 6Alibaba	149.3	–	62.0	–
49	Grok 4.3xAI	149.2	–	57.8	–
50	GPT-5.4 miniOpenAI	149.0	GPT-5.4 Mini	54.8	–
51	Kimi K2.5Moonshot AIopen	148.0	Kimi K2.5	56.1	–
52	Qwen3.6 PlusAlibabaopen	147.7	Qwen 3.6 Plus	59.0	–
53	o3-proOpenAI	147.4	–	54.1	–
54	Qwen3.7 PlusAlibaba	147.4	–	59.7	–
55	Qwen3.5 397B-A17BAlibabaopen	147.0	–	58.6	–
56	o3OpenAI	146.9	o3	60.3	–
57	Claude Sonnet 4.5Anthropic	146.8	Claude Sonnet 4.5	52.3	–
58	Qwen3.5 PlusAlibaba	146.8	–	–	–
59	MiniMax-M2.5MiniMaxopen	146.5	MiniMax-M2.5	53.5	–
60	Qwen3.6 27BAlibabaopen	146.5	Qwen3.6 27B	53.2	–

Show 100 more