Epoch Capabilities Index leaderboard: who is #1
Epoch AI · ECI, higher is better · 100 models · published Sep 24, 2026
Top 25 of 99 models, best variant per model as published
| # | Model | Score | Correct | |
|---|---|---|---|---|
| 1 |  GPT-6 AstraOpenAI | 166.6 | ||
| 2 |  Claude Fable 5.1Anthropic | 165.0 | ||
| 3 | Claude Fable 5Anthropic | 163.6 | ||
| 4 | Claude Opus 5Anthropic | 162.7 | ||
| 5 | GPT-5.5 ProOpenAI | 162.4 | ||
| 6 | GPT-5.6 SolOpenAI | 162.0 | ||
| 7 | GPT-5.6 TerraOpenAI | 159.3 | ||
| 8 | GPT-5.5OpenAI | 159.3 | ||
| 9 | GPT-5.4 ProOpenAI | 159.1 | ||
| 10 | Claude Opus 4.8Anthropic | 158.3 | ||
| 11 |  Gemini 3.7 FlashGoogle | 157.7 | ||
| 12 |  Kimi K3Moonshot AI | 157.7 | ||
| 13 | Gemini 3.8 FlashGoogle | 157.1 | ||
| 14 | GPT-5.4OpenAI | 156.9 | ||
| 15 |  Muse Spark 1.3Meta | 156.9 | ||
| 16 | GPT-5.3 CodexOpenAI | 156.8 | ||
| 17 |  Qwen3.8 Max (0902)as “Qwen 3.8 Max” | 156.7 | ||
| 18 |  Grok 4.6xAI | 156.5 | ||
| 19 | Claude Opus 4.7Anthropic | 156.4 | ||
| 20 | Claude Sonnet 5Anthropic | 156.3 | ||
| 21 | GPT-5.6 LunaOpenAI | 156.3 | ||
| 22 |  GLM-5.3Z.ai (Zhipu) | 155.6 | ||
| 23 | GPT-5.2 ProOpenAI | 155.4 | ||
| 24 |  DeepSeek V4 Pro 0813DeepSeek | 155.4 | ||
| 25 | Claude Opus 4.6Anthropic | 155.4 |
About this board and its history
What it measures
One capability score stitched together from dozens of benchmarks with item response theory.
Other reasoning boards
Source
Scores as published by Epoch AI on the capture date (2026-09-24). Source: epoch.ai eci_scores.csv (CC BY).
Sources: each leaderboard's own published results page, captured every day at 04:30 UTC. Scores exactly as each board published them; we run no evaluations. Logos via logo.dev; trademarks belong to their owners.