evals.report
BenchmarksLabsCompareRun guides
BenchmarksReasoning

Epoch Capabilities Index

A composite capability index from Epoch AI that statistically stitches together scores from 40+ benchmarks (using an Item-Response-Theory-style model) into a single saturation-resistant general-capability scale, calibrated so Claude 3.5 Sonnet=130 and GPT-5=150.

ReasoningIndexHigher is better
ScoresAboutRun this benchmark
ModelLabScore↓Source modelStatusDate
GPT-5.5 ProOpenAI159.3—OfficialApr 23, 2026Details
GPT-5.5OpenAI158.2—OfficialApr 23, 2026Details
GPT-5.4 ProOpenAI157.7—OfficialMar 5, 2026Details
Gemini 3.1 Pro PreviewGoogle DeepMind156.6—OfficialFeb 19, 2026Details
Gemini 3.5 FlashGoogle DeepMind156.3—OfficialMay 19, 2026Details
Claude Opus 4.7Anthropic156.2—OfficialApr 16, 2026Details
GPT-5.4OpenAI156.1—OfficialMar 5, 2026Details
GPT-5.3-CodexOpenAI155.9—OfficialFeb 5, 2026Details
Claude Opus 4.6Anthropic155.3—OfficialFeb 5, 2026Details
Muse SparkMeta155.1—OfficialApr 8, 2026Details
Grok 4.20 beta reasoningxAI154.3—OfficialMar 9, 2026Details
GPT-5.2OpenAI153.7—OfficialDec 11, 2025Details
Gemini 3 ProGoogle DeepMind153.4—OfficialNov 18, 2025Details
Claude Sonnet 4.6Anthropic152.6—OfficialFeb 17, 2026Details
Kimi K2.6Moonshot AI151.6—OfficialApr 20, 2026Details
Gemini 3 FlashGoogle DeepMind150.9—OfficialDec 17, 2025Details
Qwen 3.6 Max PreviewAlibaba / Qwen150.2—OfficialApr 20, 2026Details
GPT-5OpenAI150.0—OfficialAug 7, 2025Details
GLM-5.1Z.ai149.9—OfficialApr 7, 2026Details
Claude Opus 4.5Anthropic149.9—OfficialNov 24, 2025Details
GPT-5.1OpenAI149.7—OfficialNov 12, 2025Details
Qwen 3.6 PlusAlibaba / Qwen149.1—OfficialApr 2, 2026Details
Kimi K2.5Moonshot AI148.2—OfficialJan 27, 2026Details
OpenAI o3-proOpenAI148.1—OfficialJun 10, 2025Details
MiniMax M2.5MiniMax147.4—OfficialFeb 12, 2026Details
Grok 4xAI147.4—OfficialJul 9, 2025Details
o3OpenAI147.3—OfficialApr 16, 2025Details
Claude Sonnet 4.5Anthropic147.2—OfficialSep 29, 2025Details
o4-miniOpenAI146.9—OfficialApr 16, 2025Details
Gemini 2.5 ProGoogle DeepMind146.7—OfficialMar 25, 2025Details
GLM-5Z.ai146.6—OfficialFeb 11, 2026Details
DeepSeek V3.2DeepSeek146.5—OfficialDec 1, 2025Details
Qwen3.5-397B-A17BAlibaba / Qwen146.1—OfficialFeb 16, 2026Details
Kimi K2 ThinkingMoonshot AI145.6—OfficialNov 6, 2025Details
GPT-5 miniOpenAI145.6—OfficialAug 7, 2025Details
Qwen3 MaxAlibaba / Qwen144.9—OfficialSep 5, 2025Details
Claude Opus 4.1Anthropic144.7—OfficialAug 5, 2025Details
GLM-4.7Z.ai144.6—OfficialDec 22, 2025Details
Gemini 2.5 FlashGoogle DeepMind143.4—OfficialApr 17, 2025Details
Claude Opus 4Anthropic143.4—OfficialMay 22, 2025Details
Claude Haiku 4.5Anthropic143.1—OfficialOct 15, 2025Details
Claude Sonnet 4Anthropic142.6—OfficialMay 22, 2025Details
DeepSeek R1DeepSeek142.2—OfficialJan 20, 2025Details
Claude 3.7 SonnetAnthropic142.0—OfficialFeb 24, 2025Details
GLM-4.6Z.ai141.4—OfficialSep 30, 2025Details
Kimi K2 InstructMoonshot AI141.2—OfficialJul 11, 2025Details
GPT-OSS-120BOpenAI140.8—OfficialAug 5, 2025Details
Qwen3 235B A22B Instruct 2507Alibaba / Qwen139.1—OfficialJul 21, 2025Details
DeepSeek V3.1DeepSeek138.9—OfficialAug 21, 2025Details
GPT-4.1OpenAI137.6—OfficialApr 14, 2025Details
DeepSeek V3 0324DeepSeek137.2—OfficialMar 24, 2025Details
Gemini 2.0 FlashGoogle DeepMind135.9—OfficialDec 11, 2024Details
Claude 3.5 SonnetAnthropic134.4—OfficialJun 20, 2024Details
Llama 4 MaverickMeta133.1—OfficialApr 5, 2025Details
DeepSeek V3DeepSeek133.1—OfficialDec 26, 2024Details
Gemini 1.5 ProGoogle DeepMind132.8—OfficialFeb 15, 2024Details
Llama 4 ScoutMeta130.6—OfficialApr 5, 2025Details
GPT-4oOpenAI129.4—OfficialMay 13, 2024Details
Llama 3.1 405BMeta129.1—OfficialJul 23, 2024Details
Mistral LargeMistral AI121.0—OfficialFeb 26, 2024Details

Each row reports the model’s Index on Epoch Capabilities Index. Click a row for the full run context.