Terminal-Bench Science — Physical Sciences (Vals)
Science · 2026-09-23
Seventeen physical-science tasks under Vals' Terminus 2 protocol with infrastructure/provider-error retries.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Sonnet 5.5 | 58.8 |
| GPT-6 Astra | 52.9 |
| Kimi K3.1 Pro | 47.1 |
| Gemini 4 Argon | 41.2 |
| GPT-6.1 Sol | 41.2 |
| Claude Fable 5.1 | 41.2 |
| Claude Opus 5 | 35.3 |
| GPT-6 Sol | 35.3 |
| Muse Wonder Full | 35.3 |
| Claude Opus 5.5 | 35.3 |
| GPT-5.6 Sol | 29.4 |
| Muse Spark 1.3 | 29.4 |
| Grok 4.7 | 23.5 |
| Muse Wonder | 23.5 |
| Opus 4.8 | 17.6 |
| Gemini 3.7 Flash | 17.6 |
| GPT-5.6 Terra | 17.6 |
| Claude Haiku 5.5 | 11.8 |
| Sonnet 5 | 11.8 |
| DeepSeek V4.1 Flash | 11.8 |
| DeepSeek-V4-Flash-0731 | 11.8 |
| Gemini 3.8 Flash | 11.8 |
| GPT-6 Luna | 11.8 |
| Grok 4.6 | 11.8 |
| Step 5 Preview | 11.8 |
| Claude Fable 5 | 11.8 |
| DeepSeek V4 Pro 0813 | 5.9 |
| Ember-1 | 5.9 |
| Gemini 3.1 Pro Preview | 5.9 |
| Gemini 3.5 Flash | 5.9 |
| GLM-5.3 Flash | 5.9 |
| GLM-5.3 | 5.9 |
| Hy4 Preview | 5.9 |
| MiMo-V2.6-Flash | 5.9 |
| Beam 1.0 501B A23B | 0.0 |
| Gemini 3.6 Flash | 0.0 |
| GPT-5.6 Luna | 0.0 |
| Kimi K3 | 0.0 |
| Mercury 2.5 | 0.0 |
| MiMo-V2.6-Pro | 0.0 |