Terminal-Bench 2.1 - Hard
Agents · 2026-05-06
The Hard split of Terminal-Bench 2.1 reports the percentage of hard terminal-use tasks completed successfully; higher scores are better.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 87.8 |
| GPT-5.6 Sol | 84.4 |
| GPT-6 Sol | 78.9 |
| MiMo-V2.6-Flash | 77.8 |
| Kimi K3 | 74.4 |
| Gemini 3.8 Flash | 72.2 |
| Muse Spark 1.3 | 72.2 |
| GPT-5.6 Terra | 71.1 |
| GPT-5.6 Luna | 70.0 |
| Grok 4.6 | 68.9 |
| Grok 4.7 | 67.8 |
| GPT-5.5 | 66.7 |
| GPT-6 Luna | 65.6 |
| Gemini 3.6 Flash | 64.4 |
| Gemini 3.7 Flash | 64.4 |
| Opus 4.8 | 63.3 |
| MiMo-V2.6-Pro | 63.3 |
| Gemini 3.1 Pro Preview | 60.0 |
| Gemini 3.5 Flash | 60.0 |
| Sonnet 5 | 57.8 |
| DeepSeek V4.1 Flash | 56.7 |
| Grok 4.5 | 56.7 |
| Gemini 4 Argon | 53.3 |
| GLM-5.3 | 53.3 |
| Qwen3.8-Max | 53.3 |
| Muse Spark 1.1 | 52.2 |
| GLM-5.2 | 51.1 |
| Muse Spark 1.2 | 51.1 |
| Kimi K2.7 Code | 45.6 |
| Qwen3.7-Max (source unspecified) | 44.4 |
| Qwen3.8 27B | 43.3 |
| DeepSeek-V4-Flash-0731 | 41.1 |
| Opus 4.7 | 40.0 |
| Composer 2.5 | 40.0 |
| Gemini 3 Flash Preview | 40.0 |
| Hy4 Preview | 40.0 |
| MiMo-V2.5-Pro | 40.0 |
| GLM-5.3 Flash | 38.9 |
| GLM-5.1 | 37.8 |
| MiMo-V2.5 | 36.7 |