评测来源/Terminal-Bench Science(Vals 复跑)
参与排名推理
Terminal-Bench Science(Vals 复跑)
Vals AI · Terminal-Bench Science(Vals 复跑) 的公开模型评测结果。
本来源的模型成绩
| GPT-6 Astra | 62.86 | 89.4 | 2026-11-05 |
| GPT-6.1 Sol | 52.86 | 80.9 | 2026-11-05 |
| Claude Opus 5.5 | 47.14 | 76.1 | 2026-11-05 |
| Claude Sonnet 5.5 | 47.14 | 76.1 | 2026-11-05 |
| Gemini 4 Argon | 44.29 | 73.7 | 2026-11-05 |
| Claude Fable 5.1 | 40 | 70.0 | 2026-11-05 |
| GPT-6 Sol | 30 | 61.6 | 2026-11-05 |
| Claude Opus 5 | 27.14 | 59.1 | 2026-11-05 |
| GPT-5.6 Sol | 20 | 53.1 | 2026-11-05 |
| Claude Fable 5 | 15.71 | 49.4 | 2026-11-05 |
| Qwen3.8 Max | 12.86 | 47.0 | 2026-11-05 |
| Grok 4.6 | 11.43 | 45.8 | 2026-11-05 |
| GPT-5.6 Terra | 10 | 44.6 | 2026-11-05 |
| Muse Spark 1.3 | 10 | 44.6 | 2026-11-05 |
| Grok 4.7 | 10 | 44.6 | 2026-11-05 |
| Claude Haiku 5.5 | 10 | 44.6 | 2026-11-05 |
| Gemini 3.8 Flash | 8.57 | 43.4 | 2026-11-05 |
| Claude Sonnet 5 | 5.71 | 41.0 | 2026-11-05 |
| GLM 5.3 | 5.71 | 41.0 | 2026-11-05 |
| MiMo V2.6 Flash | 4.29 | 39.8 | 2026-11-05 |
| Claude Opus 4.8 | 4.29 | 39.8 | 2026-11-05 |
| Gemini 3.6 Flash | 4.29 | 39.8 | 2026-11-05 |
| DeepSeek V4 Flash 0731 | 4.29 | 39.8 | 2026-11-05 |
| DeepSeek V4 Pro 0813 | 4.29 | 39.8 | 2026-11-05 |
| GPT-6 Luna | 4.29 | 39.8 | 2026-11-05 |
| GLM-5.3 Flash | 2.86 | 38.6 | 2026-11-05 |
| MiMo V2.6 Pro | 2.86 | 38.6 | 2026-11-05 |
| Step 5 Preview | 2.86 | 38.6 | 2026-11-05 |
| Kimi K3 | 1.43 | 37.3 | 2026-11-05 |
| GPT-5.6 Luna | 0 | 36.1 | 2026-11-05 |
| DeepSeek V4.1 Flash | 0 | 36.1 | 2026-11-05 |