Pass IndexThe State of AISign in

Gbaeval — Epoch AI

Gbaeval — Epoch AI is run by Epoch AI. It has ranked 23 models, of which the catalogue holds 23, scoring from 0 to 0.796.

measured by Epoch AI · the board itself

PlaceModelMetricScore
1stof 23Claude Opus 5Overall score0.796
2ndof 23Claude Fable 5Overall score0.745
3rdof 23Claude Opus 4.8Overall score0.709
4thof 23Grok 4.5Overall score0.654
5thof 23Claude Sonnet 5Overall score0.653
6thof 23GPT-5.5Overall score0.532
7thof 23GPT-5.6 SolOverall score0.526
8thof 23Claude Sonnet 4.6Overall score0.488
9thof 23Kimi K3Overall score0.483
10thof 23GPT-5.4Overall score0.451
11thof 23Claude Opus 4.6Overall score0.441
12thof 23Claude Opus 4.7Overall score0.438
13thof 23Muse Spark 1.1Overall score0.079
14thof 23Gemini 3.5 FlashOverall score0.067
15thof 23Grok Build 0.1Overall score0.024
16thof 23MiniMax M3Overall score0.009
17thof 23Kimi K2.6Overall score0.009
18thof 23Kimi K2.7 CodeOverall score0.009
19thof 23Gemini 3.1 Pro PreviewOverall score0.008
20thof 23Qwen3.7 MaxOverall score (max)0.004
21stof 23GLM 5.2Overall score0
22ndof 23MiniMax M2.7Overall score0
23rdof 23GLM 5.1Overall score0

Read from the board on 2026-09-16