MatFlowBench Computational Pass@2
Science · 2026-10-09
Share of the 25 computational MatFlowBench tasks that a model passes in at least one of its two rollouts.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 72.0 |
| Claude Opus 5.5 | 68.0 |
| Claude Sonnet 5.5 | 68.0 |
| GPT-6.1 Sol | 68.0 |
| Kimi K3 | 28.0 |
| Grok 4.7 | 24.0 |
| Hy4 Preview | 20.0 |
| Muse Spark 1.3 | 12.0 |
| Gemini 3.8 Flash | 4.0 |