Benchmarks
Surface Evolver Bench
Models scored
17
Best score
0.95
Metric
Mean score
Surface Evolver Bench is an AI evaluation tracked by Epoch AI, with 17 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.
Scored models span August 2025 to July 2026. The best score recorded is 0.95, measured as Mean score. Models come from Anthropic, OpenAI, Moonshot, xAI, Z.ai (Zhipu AI), DeepSeek, MiniMax, Alibaba and others.
Full record
- Score metric
- Mean score
- Models scored
- 17
- Best score recorded
- 0.95
- Earliest model scored
- Aug 5, 2025
- Most recent model scored
- Jul 16, 2026
- Third-party reported
- Yes
Leaderboard
| 01 | claude-fable-5_high | Anthropic | 0.95 |
| 02 | gpt-5.6-sol_xhigh | OpenAI | 0.93 |
| 03 | kimi-k3_max | Moonshot | 0.93 |
| 04 | gpt-5.5_high | OpenAI | 0.88 |
| 05 | claude-opus-4-8_high | Anthropic | 0.88 |
| 06 | gpt-5.6-terra_xhigh | OpenAI | 0.84 |
| 07 | gpt-5.5_medium | OpenAI | 0.81 |
| 08 | grok-4.5_high | xAI | 0.74 |
| 09 | claude-opus-4-8_none | Anthropic | 0.66 |
| 10 | gpt-5.6-luna_medium | OpenAI | 0.6 |
| 11 | glm-5.2_high | Z.ai (Zhipu AI) | 0.56 |
| 12 | MiniMax-M3 | MiniMax | 0.53 |
| 13 | kimi-k2.7-code | Moonshot | 0.49 |
| 14 | qwen3.6-35b-a3b | Alibaba | 0.44 |
| 15 | deepseek-v4-pro_high | DeepSeek | 0.4 |
| 16 | gemma-4-31b-it | Google DeepMind | 0.31 |
| 17 | gpt-oss-120b | OpenAI | 0.25 |
More benchmarks
SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.
← All benchmarks