Surface Evolver Bench Pass Rate: leaderboard
Surface Evolver Bench pass rate measures fully passing agentic submissions for custom liquid-surface physics simulations.
Metric: Pass Rate (%). Source: yhenon.github.io. Saturation forecast: Around January 2027. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K3 | 87.5 | #30 |
| 2 | GPT-5.6 Sol (xHigh) | 87.5 | #19 (GPT-5.6 Sol) |
| 3 | Claude Fable 5 (High) | 87.5 | #16 (Claude Fable 5) |
| 4 | GPT-5.5 (High) | 75 | #31 (GPT-5.5) |
| 5 | GPT-5.6 Terra (xHigh) | 68.75 | #41 (GPT-5.6 Terra) |
| 6 | Claude Opus 4.8 (High) | 68.75 | #37 (Claude Opus 4.8) |
| 7 | GPT-5.5 (Medium) | 62.5 | #31 (GPT-5.5) |
| 8 | Gemini 3.8 Flash (High) | 56.25 | #21 (Gemini 3.8 Flash) |
| 9 | Grok 4.5 (High) | 50 | #48 (Grok 4.5) |
| 10 | Kimi K2.7 Code | 37.5 | #89 |
| 11 | Gemini 3.5 Flash (Medium) | 37.5 | #60 (Gemini 3.5 Flash) |
| 12 | Muse Spark 1.1 (High) | 37.5 | #52 (Muse Spark 1.1) |
| 13 | GLM-5.2 (High) | 31.25 | #73 (GLM-5.2) |
| 14 | DeepSeek V4.1 Flash (High) | 31.25 | #36 (DeepSeek V4.1 Flash) |
| 15 | Claude Opus 4.8 | 25 | #37 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=surface-evolver-bench-pass-rate · How It Works · Data refreshed daily, snapshot 2026-10-20.