OpenAI ECI: leaderboard
OpenAI's own capabilities index, built the way Epoch builds its ECI but fitted by OpenAI over public benchmarks plus its internal ones, so one number places a model against all of them. Covers 48 public models from OpenAI and other labs, each with a 90% bootstrap interval; OpenAI notes its own models score higher here than on Epoch's index. Higher is better. Source: OpenAI, Research acceleration: The jagged frontier (November 2026).
Metric: OpenAI ECI score. Source: openai.com. Saturation forecast: Around February 2028. 48 models tracked. Updated 22 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6.2 Astra | 170.3 | #4 |
| 2 | Claude Fable 5.2 | 169.7 | #2 |
| 3 | GPT-6.2 Sol | 169.1 | #10 |
| 4 | GPT-6 Astra | 168.9 | #11 |
| 5 | GPT-6.1 Sol | 165.7 | #15 |
| 6 | Claude Opus 5.5 | 165.2 | #9 |
| 7 | Gemini 4 Argon | 163.2 | #18 |
| 8 | Claude Fable 5.1 | 162.5 | #20 |
| 9 | Claude Sonnet 5.5 | 162.4 | #24 |
| 10 | GPT-5.6 Sol | 162.1 | #36 |
| 11 | GPT-6 Sol | 161.7 | #33 |
| 12 | Claude Opus 5 | 160.3 | #26 |
| 13 | Claude Fable 5 | 160.2 | #32 |
| 14 | GPT-5.5 | 159.8 | #56 |
| 15 | Muse Wonder | 159.7 | #34 |
No result here: #3 Gemini 4 Neon, #7 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=openai-eci · How It Works · Data refreshed daily, snapshot 2026-11-22.