Valumigo

Epoch AI · Overall capability index (ECI)

We regularly fetch and display public benchmark data: LMArena user-voted rankings and Epoch AI test scores. Each leaderboard shows its publication and retrieval dates. These scores are not produced by this site.

LMArena scores come from people comparing two models' answers side by side and voting for the better one, using an Elo-based system. If score differences fall within the confidence intervals, treat the models as roughly comparable.

How to read this leaderboard

What it measures: A composite capability index from Epoch AI that combines multiple benchmark scores on a common scale. It statistically estimates each benchmark's difficulty and how its scores change, enabling comparisons between models evaluated on different tests.

How to read the score: This is an estimate of relative capability, not an accuracy rate. Score differences cannot be directly converted into differences you would experience in actual use. Use it as a reference for overall standing across multiple tests.

Caveats: For capabilities needed for a specific use, also check the relevant benchmarks individually. Estimates and uncertainty may change as evaluation results or included benchmarks are added.

Insights from this ranking

  • The highest score is 167.4, achieved by Claude Opus 5.5.
  • The gap between 1st and 5th place is 4.5.
  • Anthropic has the most models among the top 10, with 5.
  • Results for this test are public for 60 models.

Overall capability index (ECI) Epoch AI

ECI combines results from multiple tests into an overall index created by Epoch AI. Higher scores indicate stronger overall capabilities. · Retrieved 2026-10-04

156159162165168
AnthropicClaude Opus 5.5
167.4
OGPT-6 Astra
166.5
AnthropicClaude Sonnet 5.5
165.2
AnthropicClaude Fable 5.1
164.8
AnthropicClaude Opus 5
162.9
OGPT-5.5 Pro
162.4
AnthropicClaude Fable 5
162.2
OGPT-5.6 Sol
161.8
OGPT-5.6 Terra
159.8
OGPT-5.5
159.2
OGPT-5.4 Pro
159.1
AnthropicClaude Opus 4.8
158.3
Moonshot AIKimi K3
157.6
GoogleGemini 3.7 Flash
157.4
GoogleGemini 3.8 Flash
156.9
OGPT-5.4
156.9
MetaMuse Spark 1.3
156.9
OGPT-5.3 Codex
156.8
xGrok 4.6
156.6
AlibabaQwen 3.8 Max
156.6
View table (top 40)
RankModelDeveloperRelease dateScore
1AnthropicClaude Opus 5.5Anthropic2026-09-22167.4
2OGPT-6 AstraOpenAI2026-09-03166.5
3AnthropicClaude Sonnet 5.5Anthropic2026-09-28165.2
4AnthropicClaude Fable 5.1Anthropic2026-09-01164.8
5AnthropicClaude Opus 5Anthropic2026-07-24162.9
6OGPT-5.5 ProOpenAI2026-04-23162.4
7AnthropicClaude Fable 5Anthropic2026-06-09162.2
8OGPT-5.6 SolOpenAI2026-07-09161.8
9OGPT-5.6 TerraOpenAI2026-07-09159.8
10OGPT-5.5OpenAI2026-04-23159.2
11OGPT-5.4 ProOpenAI2026-03-05159.1
12AnthropicClaude Opus 4.8Anthropic2026-05-28158.3
13Moonshot AIKimi K3Moonshot AI2026-07-16157.6
14GoogleGemini 3.7 FlashGoogle2026-08-13157.4
15GoogleGemini 3.8 FlashGoogle2026-09-02156.9
16OGPT-5.4OpenAI2026-03-05156.9
17MetaMuse Spark 1.3Meta2026-09-02156.9
18OGPT-5.3 CodexOpenAI2026-02-05156.8
19xGrok 4.6xAI2026-08-12156.6
20AlibabaQwen 3.8 MaxAlibaba2026-08-02156.6
21OGPT-5.6 LunaOpenAI2026-07-09156.5
22AnthropicClaude Opus 4.7Anthropic2026-04-16156.4
23AnthropicClaude Sonnet 5Anthropic2026-06-30156.3
24ZGLM-5.3Z.ai2026-08-14155.8
25OGPT-5.2 ProOpenAI2025-12-11155.4
26DeepSeekDeepSeek V4 Pro 0813DeepSeek2026-08-13155.4
27AnthropicClaude Opus 4.6Anthropic2026-02-05155.3
28AlibabaQwen3.8 Max (0902)Alibaba2026-09-01155.2
29DeepSeekDeepSeek V4.1 FlashDeepSeek2026-09-09155
30MetaMuse Spark 1.2Meta2026-08-05155
31GoogleGemini 3.1 ProGoogle2026-02-19154.8
32DeepSeekDeepSeek V4 Flash 0731DeepSeek2026-07-31154.5
33GoogleGemini 3.5 FlashGoogle2026-05-19154.5
34GoogleGemini 3.6 FlashGoogle2026-07-21154.3
35MetaMuse Spark 1.1Meta2026-07-09154.3
36xGrok 4.5xAI2026-07-08154
37AlibabaQwen3.7-MaxAlibaba2026-05-19153.7
38OGPT-5.2OpenAI2025-12-11153.5
39GoogleGemini 3 ProGoogle2025-11-18153
40AnthropicClaude Sonnet 4.6Anthropic2026-02-17152.3

Data: LMArena leaderboard dataset (CC BY 4.0). Changes: Top 50 entries per leaderboard; scores are rounded. https://huggingface.co/datasets/lmarena-ai/leaderboard-datasetCompany logos are trademarks of their respective owners and are used only for identification (icons: Simple Icons).Data: Epoch AI — Capabilities & benchmarking (CC BY 4.0). https://epoch.ai/benchmarks