End-to-end terminal tasks on Terminal-Bench 2
49.6ScoreGPT-5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5Model Type=Proprietary2025.12 | 49.6 | |
| Claude-Sonnet-4.5Model Type=Proprietary2025.12 | 42.8 | |
| Kimi-K2-thinkingModel Type=Open Source, Parameter Scale=> 100B2025.12 | 35.7 | |
| Gemini-2.5-proModel Type=Proprietary2025.12 | 32.6 | |
| DeepSeek-V3.1-Nex-N1Model Type=Open Source, Parameter Scale=> 100B2025.12 | 31.8 | |
| Minimax-M2Model Type=Open Source, Parameter Scale=> 100B2025.12 | 30 | |
| GLM-4.6Model Type=Open Source, Parameter Scale=> 100B2025.12 | 24.5 | |
| DeepSeek-V3.1Model Type=Open Source, Parameter Scale=> 100B2025.12 | 22.2 | |
| Qwen3-32B-Nex-N1Model Type=Open Source, Parameter Scale=< 100B2025.12 | 16.7 | |
| Qwen3-30B-A3B-Nex-N1Model Type=Open Source, Parameter Scale=< 100B2025.12 | 8.3 | |
| Qwen3-32BModel Type=Open Source, Parameter Scale=< 100B2025.12 | 7.9 | |
| Qwen3-30B-A3BModel Type=Open Source, Parameter Scale=< 100B2025.12 | 6 | |
| InternLM3-8B-Nex-N1Model Type=Open Source, Parameter Scale=< 100B2025.12 | 0 |