Terminal-Bench 4.0 — Software
Agents · 2026-09-22
Resolution rate on the 18 software tasks in Terminal-Bench 4.0.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 4 Carbon | 63.0 |
| GPT-6 Astra | 61.1 |
| GPT-6.1 Sol | 57.4 |
| Muse Wonder | 57.4 |
| GPT-6 Sol | 53.7 |
| Kimi K3.1 Pro | 53.7 |
| Muse Wonder Full | 53.7 |
| GPT-6.1 Terra | 50.0 |
| Gemini 4 Argon | 48.1 |
| Claude Haiku 5.5 | 46.3 |
| GPT-5.6 Sol | 46.3 |
| Qwen3.8 Max (0902) | 44.4 |
| MiMo-V2.6-Pro | 38.9 |
| DeepSeek V4.1 Flash 1017 | 37.0 |
| GLM-5.3 Flash | 35.2 |
| GLM-5.3 | 35.2 |
| GPT-5.6 Terra | 35.2 |
| Mistral Large 4 Preview | 35.2 |
| Step 5 Preview | 35.2 |
| DeepSeek V4.1 Flash | 33.3 |
| Grok 4.7 | 29.6 |
| MiMo-V2.6-Flash | 29.6 |
| Muse Spark 1.3 | 29.6 |
| Ember-1 | 25.9 |
| Grok 4.6 | 25.9 |
| DeepSeek-V4-Flash-0731 | 24.1 |
| DeepSeek V4 Pro 0813 | 22.2 |
| Kimi K3 | 22.2 |
| MiniMax M3.1 Flash | 22.2 |
| Gemini 3.8 Flash | 18.5 |
| Opus 4.8 | 16.7 |
| Sonnet 5 | 14.8 |
| GPT-5.6 Luna | 14.8 |
| GPT-6 Luna | 14.8 |
| Hy4 Preview | 14.8 |
| Beam 1.0 501B A23B | 11.1 |
| Muse Spark 1.2 | 11.1 |
| Muse Spark Open | 11.1 |
| Gemini 3.7 Flash | 7.4 |
| GPT-5.4 Mini | 7.4 |