Skip to content

Benchmark

Surface Evolver Bench Pass rate

Scores are percentages. Higher is better.

Models measured
26
Scores recorded
28
Sources
1
IQ weight
1.64

Score distribution

36.6%
Mean
31.3%
Median
26.7%
Std dev
6.3%
10th percentile
75%
90th percentile
6.3%
Min
6.3%87.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.64 for chat models, measured over 26 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude Fable 5high· Epoch AI Benchmarking Hub data87.5%
2GPT-5.6 Solxhigh· Epoch AI Benchmarking Hub data87.5%
3Kimi K3· Epoch AI Benchmarking Hub data87.5%
4GPT 5.5high· Epoch AI Benchmarking Hub data75%
5Claude Opus 4.8high· Epoch AI Benchmarking Hub data68.8%
6GPT-5.6 Terraxhigh· Epoch AI Benchmarking Hub data68.8%
7GPT 5.5medium· Epoch AI Benchmarking Hub data62.5%
8Gemini 3.8 Flashhigh· Epoch AI Benchmarking Hub data56.3%
9Grok 4.5high· Epoch AI Benchmarking Hub data50%
10Gemini 3.5 Flashmedium· Epoch AI Benchmarking Hub data37.5%
11Kimi K2.7 Code· Epoch AI Benchmarking Hub data37.5%
12Muse Spark 1.1high· Epoch AI Benchmarking Hub data37.5%
13DeepSeek V4.1 Flashhigh· Epoch AI Benchmarking Hub data31.3%
14GLM 5.2high· Epoch AI Benchmarking Hub data31.3%
15Claude Opus 4.8none· Epoch AI Benchmarking Hub data25%
16DeepSeek V4 Prohigh· Epoch AI Benchmarking Hub data25%
17Claude Sonnet 5medium· Epoch AI Benchmarking Hub data18.8%
18GLM 5.3 Flashmax· Epoch AI Benchmarking Hub data18.8%
19GPT-5.6 Lunamedium· Epoch AI Benchmarking Hub data18.8%
20MiniMax M3· Epoch AI Benchmarking Hub data18.8%
21Mistral Large 4· Epoch AI Benchmarking Hub data18.8%
22Qwen 3.8 27Bxhigh· Epoch AI Benchmarking Hub data18.8%
23Qwen3.6 35B A3B· Epoch AI Benchmarking Hub data12.5%
24Gemma 4 31B IT· Epoch AI Benchmarking Hub data6.3%
25GPT OSS 120B· Epoch AI Benchmarking Hub data6.3%
26Laguna M.1· Epoch AI Benchmarking Hub data6.3%
27Mistral Medium2604· Epoch AI Benchmarking Hub data6.3%
28Trinity Large Thinking· Epoch AI Benchmarking Hub data6.3%

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)