Surface evolver bench — Epoch AI

Surface evolver bench — Epoch AI is run by Epoch AI. It has ranked 29 models, of which the catalogue holds 24, scoring from 0.156 to 0.95.

measured by Epoch AI · the board itself

Place	Model	Metric	Score
1stof 29	Claude Fable 5	Mean score (high)	0.95
2ndof 29	Kimi K3	Mean score	0.95
3rdof 29	GPT-5.6 Sol	Mean score (xhigh)	0.931
5thof 29	GPT-5.5	Mean score (high)	0.881
6thof 29	Claude Opus 4.8	Mean score (high)	0.875
7thof 29	GPT-5.6 Terra	Mean score (xhigh)	0.838
9thof 29	Gemini 3.8 Flash	Mean score (high)	0.769
10thof 29	Grok 4.5	Mean score (high)	0.744
12thof 29	GPT-5.6 Luna	Mean score (medium)	0.619
13thof 29	Claude Sonnet 5	Mean score (medium)	0.6
14thof 29	Gemini 3.5 Flash	Mean score (medium)	0.581
15thof 29	GLM 5.2	Mean score (high)	0.556
16thof 29	MiniMax M3	Mean score	0.55
17thof 29	Muse Spark 1.1	Mean score (high)	0.525
18thof 29	GLM 5.3 Flash	Mean score (max)	0.525
19thof 29	Kimi K2.7 Code	Mean score	0.487
20thof 29	DeepSeek V4.1 Flash	Mean score (high)	0.463
21stof 29	Qwen3.8 27B	Mean score (xhigh)	0.45
22ndof 29	Qwen3.6 35B A3B	Mean score	0.444
23rdof 29	DeepSeek V4 Pro	Mean score (high)	0.4
24thof 29	Gemma 4 31B	Mean score	0.306
27thof 29	GPT OSS 120B	Mean score	0.25
28thof 29	Laguna M.1	Mean score	0.156
29thof 29	Trinity Large Thinking	Mean score	0.156

Place

Model

Metric

Score

1stof 29

Mean score (high)

0.95

2ndof 29

Mean score

0.95