Epoch Capabilities Index
Epoch AI's own composite capability score fitted across many benchmarks. Shown for reference; not part of the BenchLeader Index.
As of 24 Oct 2026, Claude Opus 5.5 leads Epoch Capabilities Index on BenchLeader with 168.8, ahead of GPT-6 Astra at 168.8, across 245 model configurations with a published result.
- Published by
- Epoch AI Benchmarking Hub
- Category
- Composite
- Index weight
- Reference only
- Models
- 245
- Data as of
- 24 Oct 2026
- Newest result
- 15 Oct 2026
CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.
What the test looks like
The Epoch Capabilities Index is Epoch AI's own composite: a single score fitted across many benchmarks with an item-response model, so results on tests of different difficulty land on one scale.
How it is scored
A fitted capability score; higher is better and the scale is arbitrary. Shown for reference only, not part of the BenchLeader Index.
What to keep in mind
It is a model-based estimate rather than a measurement, and it is updated when Epoch refits the model, which can move every score at once. Compare it with the BenchLeader Index as a sanity check, not as an input.
- 1Claude Opus 5.5168.8
- 2GPT-6 Astra168.8
- 3GPT-6.1 Sol168.4
- 4Claude Fable 5.1166.5
- 5Gemini 4 Argon166.4
- 6Claude Sonnet 5.5166.2
- 7Claude Opus 5163.9
- 8GPT-6 Sol163.7
- 9Claude Fable 5163.4
- 10GPT-5.5 Pro162.9
- 11GPT-5.6 Sol162.8
- 12Muse Wonder161.8
- 13GPT-5.6 Terra160.1
- 14GPT-5.5159.7
- 15GPT-5.4 Pro159.6
| # | ||||
|---|---|---|---|---|
| 1 |  Claude Opus 5.5 | 168.8 | 62.0 | 2026-09-22 |
| 2 |  GPT-6 Astra | 168.8 | 68.0 | 2026-09-03 |
| 3 | GPT-6.1 Sol | 168.4 | 63.0 | 2026-09-29 |
| 4 | Claude Fable 5.1 | 166.5 | 66.4 | 2026-09-01 |
| 5 |  Gemini 4 Argon | 166.4 | 64.9 | 2026-09-30 |
| 6 | Claude Sonnet 5.5 | 166.2 | – | 2026-09-28 |
| 7 | Claude Opus 5 | 163.9 | 65.5 | 2026-07-24 |
| 8 | GPT-6 Sol | 163.7 | – | 2026-09-22 |
| 9 | Claude Fable 5 | 163.4 | 67.1 | 2026-06-09 |
| 10 | GPT-5.5 Pro | 162.9 | – | 2026-04-23 |
| 11 | GPT-5.6 Sol | 162.8 | 55.9 | 2026-07-09 |
| 12 |  Muse Wonder | 161.8 | 63.7 | 2026-10-12 |
| 13 | GPT-5.6 Terra | 160.1 | 52.7 | 2026-07-09 |
| 14 | GPT-5.5 | 159.7 | 60.3 | 2026-04-23 |
| 15 | GPT-5.4 Pro | 159.6 | 63.3 | 2026-03-05 |
| 16 | Claude Haiku 5.5 | 159.3 | – | 2026-10-07 |
| 17 | Claude Opus 4.8 | 159.0 | 60.1 | 2026-05-28 |
| 18 | Gemini 3.7 Flash | 158.0 | 53.4 | 2026-08-13 |
| 19 |  Kimi K3open ↗ | 157.9 | 59.1 | 2026-07-16 |
| 20 | Gemini 3.8 Flash | 157.4 | 55.7 | 2026-09-02 |
| 21 |  Qwen3.8 Maxmax | 157.4 | 63.3 | 2026-08-02 |
| 22 | GPT-5.4 | 157.3 | 58.2 | 2026-03-05 |
| 23 | Muse Spark 1.3 | 157.2 | 59.8 | 2026-09-02 |
| 24 | GPT-5.3-Codex | 157.1 | – | 2026-02-05 |
| 25 | Claude Sonnet 5 | 156.9 | 56.1 | 2026-06-30 |
| 26 | Claude Opus 4.7 | 156.8 | 63.3 | 2026-04-16 |
| 27 | GPT-6 Luna | 156.8 | – | 2026-09-22 |
| 28 |  Grok 4.6 | 156.6 | 57.8 | 2026-08-12 |
| 29 | GPT-5.6 Luna | 156.5 | 48.0 | 2026-07-09 |
| 30 | GPT-5.2 Pro | 156.3 | – | 2025-12-11 |
| 31 |  GLM 5.3open ↗ | 156.0 | – | 2026-08-14 |
| 32 |  DeepSeek V4.1 Flashopen ↗ | 155.9 | – | 2026-09-09 |
| 33 | Claude Opus 4.6 | 155.7 | 62.6 | 2026-02-05 |
| 34 | DeepSeek V4 Proopen ↗ | 155.6 | 53.5 | 2026-08-13 |
| 35 | Gemini 3.1 Pro | 155.3 | 60.6 | 2026-02-19 |
| 36 | Muse Spark Openopen ↗ | 155.1 | – | 2026-10-15 |
| 37 | Muse Spark 1.2 | 155.1 | – | 2026-08-05 |
| 38 | Gemini 3.5 Flash | 154.9 | 49.1 | 2026-05-19 |
| 39 | DeepSeek V4 Flashopen ↗ | 154.8 | 52.2 | 2026-07-31 |
| 40 | Gemini 3.6 Flash | 154.5 | 49.5 | 2026-07-21 |
| 41 | Grok 4.7 | 154.4 | – | 2026-09-21 |
| 42 | Grok 4.5 | 154.1 | 59.0 | 2026-07-08 |
| 43 | Muse Spark 1.1 | 154.1 | 62.5 | 2026-07-09 |
| 44 | Qwen3.7max | 154.0 | 60.6 | 2026-05-19 |
| 45 | GPT-5.2 | 153.7 | 58.0 | 2025-12-11 |
| 46 | GLM 5.3 Flashopen ↗ | 153.1 | 62.6 | 2026-08-20 |
| 47 | Gemini 3 Pro | 153.1 | 60.4 | 2025-11-18 |
| 48 | Claude Sonnet 4.6 | 152.8 | 57.1 | 2026-02-17 |
| 49 | Grok 4.20 | 152.1 | – | 2026-02-17 |
| 50 | Muse Spark | 152.0 | 64.3 | 2026-04-08 |
| 51 | GLM 5.2open ↗ | 151.9 | 52.0 | 2026-06-16 |
| 52 | Gemini 3 Flash | 151.9 | 57.0 | 2025-12-17 |
| 53 | Qwen3.8 27Bopen ↗ | 151.7 | 54.6 | 2026-08-14 |
| 54 | Kimi K2.6open ↗ | 150.9 | 55.4 | 2026-04-20 |
| 55 | GPT-5 Pro | 150.6 | 59.8 | 2025-10-07 |
| 56 | Inkling Smallopen ↗ | 150.3 | 54.8 | 2026-07-15 |
| 57 | Claude Opus 4.5 | 150.1 | 57.9 | 2025-11-24 |
| 58 | Kimi K2.7 Codeopen ↗ | 150.1 | 55.8 | 2026-06-12 |
| 59 |  Mistral Large 4open | 150.0 | 56.6 | 2026-10-06 |
| 60 | GPT-5 | 150 | 56.2 | 2025-08-07 |
More composite benchmarks · All LLM benchmarks
Cite as: BenchLeader, “Epoch Capabilities Index leaderboard”, data as of 24 Oct 2026, https://www.benchleader.com/benchmarks/epoch_eci.
Epoch Capabilities Index: questions
- What does Epoch Capabilities Index measure?
- The Epoch Capabilities Index is Epoch AI's own composite: a single score fitted across many benchmarks with an item-response model, so results on tests of different difficulty land on one scale. Scores are reported in score; higher is better.
- Which AI model leads Epoch Capabilities Index?
- Claude Opus 5.5 leads Epoch Capabilities Index with 168.8 as of 24 Oct 2026, ahead of GPT-6 Astra at 168.8.
- How many models have Epoch Capabilities Index results?
- 245 model configurations have a Epoch Capabilities Index result on BenchLeader, all taken from Epoch AI Benchmarking Hub.
- Who runs Epoch Capabilities Index and how often is it updated?
- Epoch Capabilities Index is published by Epoch AI Benchmarking Hub. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does Epoch Capabilities Index count toward the BenchLeader Index?
- No. Epoch Capabilities Index is shown for reference but left out of the composite index.