benchmarks / terminal-bench-4
Terminal-Bench 4.0
accuracy. Status active. Contamination risk low. Maintainer Terminal-Bench / Laude Institute. Source.
| Model | Score | Date | Kind |
|---|---|---|---|
| Claude Opus 5.5 | 64.85 | 2026-10-06 | snapshot |
| Claude Sonnet 5.5 | 61.82 | 2026-10-06 | snapshot |
| GPT-6 Astra | 58.18 | 2026-10-06 | snapshot |
| GPT-6.1 Sol | 58.18 | 2026-10-06 | snapshot |
| Claude Fable 5.1 | 57.88 | 2026-10-06 | snapshot |
| GPT-6 Astra | 57.88 | 2026-10-06 | snapshot |
| GPT-6 Astra | 57.88 | 2026-10-06 | snapshot |
| Claude Fable 5.1 | 57.88 | 2026-10-06 | snapshot |
| Claude Fable 5.1 | 54.55 | 2026-10-06 | snapshot |
| GPT-6 Astra | 54.24 | 2026-10-06 | snapshot |
| Claude Fable 5.1 | 53.94 | 2026-10-06 | snapshot |
| Claude Opus 5 | 53.94 | 2026-10-06 | snapshot |
| Claude Opus 5 | 51.82 | 2026-10-06 | snapshot |
| GPT-6 Astra | 50.61 | 2026-10-06 | snapshot |
| Claude Opus 5 | 50.3 | 2026-10-06 | snapshot |
| GPT-6 Sol | 49.39 | 2026-10-06 | snapshot |
| Claude Opus 5 | 44.85 | 2026-10-06 | snapshot |
| Claude Fable 5 | 44.55 | 2026-10-06 | snapshot |
| Claude Fable 5.1 | 43.33 | 2026-10-06 | snapshot |
| GLM 5.3 | 41.82 | 2026-10-06 | snapshot |
| Grok 4.7 | 37.58 | 2026-10-06 | snapshot |
| GPT-5.6 Sol | 37.27 | 2026-10-06 | snapshot |
| GLM 5.3 Flash | 35.76 | 2026-10-06 | snapshot |
| Claude Opus 5 | 34.85 | 2026-10-06 | snapshot |
| Qwen3 8 Max 0902 | 26.97 | 2026-10-06 | snapshot |
| Claude Opus 4.8 | 23.64 | 2026-10-06 | snapshot |
| GPT-5.6 Terra | 21.52 | 2026-10-06 | snapshot |
| Grok 4.6 | 20.3 | 2026-10-06 | snapshot |
| Gemini 3.8 Flash | 19.09 | 2026-10-06 | snapshot |
| GPT-5.6 Luna | 17.27 | 2026-10-06 | snapshot |