Skip to content

Benchmark

Surface Evolver Bench Mean score

Scores are percentages.

Models measured
26
Scores recorded
29
Sources
1
IQ weight
1.34

Score distribution

58.6%
Mean
55.6%
Median
24.3%
Std dev
26.9%
10th percentile
93%
90th percentile
15.6%
Min
15.6%95%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.34 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 data95%
2Kimi K3· Epoch AI Benchmarking Hub data95%
3GPT-5.6 Solxhigh· Epoch AI Benchmarking Hub data93.1%
4Kimi K3max· Epoch AI Benchmarking Hub data93%
5GPT 5.5high· Epoch AI Benchmarking Hub data88.1%
6Claude Opus 4.8high· Epoch AI Benchmarking Hub data87.5%
7GPT-5.6 Terraxhigh· Epoch AI Benchmarking Hub data83.8%
8GPT 5.5medium· Epoch AI Benchmarking Hub data81.3%
9Gemini 3.8 Flashhigh· Epoch AI Benchmarking Hub data76.9%
10Grok 4.5high· Epoch AI Benchmarking Hub data74.4%
11Claude Opus 4.8none· Epoch AI Benchmarking Hub data68.1%
12GPT-5.6 Lunamedium· Epoch AI Benchmarking Hub data61.9%
13Claude Sonnet 5medium· Epoch AI Benchmarking Hub data60%
14Gemini 3.5 Flashmedium· Epoch AI Benchmarking Hub data58.1%
15GLM 5.2high· Epoch AI Benchmarking Hub data55.6%
16MiniMax M3· Epoch AI Benchmarking Hub data55%
17GLM 5.3 Flashmax· Epoch AI Benchmarking Hub data52.5%
18Muse Spark 1.1high· Epoch AI Benchmarking Hub data52.5%
19Kimi K2.7 Code· Epoch AI Benchmarking Hub data48.8%
20DeepSeek V4.1 Flashhigh· Epoch AI Benchmarking Hub data46.3%
21Qwen 3.8 27Bxhigh· Epoch AI Benchmarking Hub data45%
22Qwen3.6 35B A3B· Epoch AI Benchmarking Hub data44.4%
23DeepSeek V4 Prohigh· Epoch AI Benchmarking Hub data40%
24Gemma 4 31B IT· Epoch AI Benchmarking Hub data30.6%
25Mistral Large 4· Epoch AI Benchmarking Hub data28.1%
26Mistral Medium2604· Epoch AI Benchmarking Hub data26.9%
27GPT OSS 120B· Epoch AI Benchmarking Hub data25%
28Laguna M.1· Epoch AI Benchmarking Hub data15.6%
29Trinity Large Thinking· Epoch AI Benchmarking Hub data15.6%

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