Code Agent Task on Terminal-Bench 2.0
82TB 2.0 ScoreGPT-5.5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5.5Size=-, Post-Training Data=-, Scaffold=Codex CLI2026.06 | 82 | |
| Claude Opus 4.6Size=-, Post-Training Data=-, Scaffold=Terminus-22026.06 | 62.9 | |
| GPT-5.2 HighAgentic Environment=Terminus22026.06 | 62.4 | |
| Claude-Opus-4.5Agentic Environment=Terminus22026.06 | 60.2 | |
| Gemini-3-Flash HighEvaluation Framework=Terminus22026.06 | 60 | |
| Gemini-3-pro HighAgentic Environment=Terminus22026.06 | 56.9 | |
| Seed2.0 ProAgentic Environment=Terminus22026.06 | 55.8 | |
| GLM 5Size=744B, Post-Training Data=-, Scaffold=Terminus-22026.06 | 52.4 | |
| Claude-Sonnet-4.5Agentic Environment=Terminus22026.06 | 45.2 | |
| Seed2.0 LiteEvaluation Framework=Terminus22026.06 | 45 | |
| DeepSeek-V3.2Size=685B, Post-Training Data=-, Scaffold=Terminus-22026.06 | 39.6 | |
| GPT-5-mini HighEvaluation Framework=Terminus22026.06 | 36.9 | |
| Nemotron-Terminal-Qwen3-32BSize=32B, Post-Training Data=490.5K, Scaffold=Terminus-22026.06 | 27.4 | |
| Terminal-Lego-Qwen3-32BSize=32B, Post-Training Data=15.3K, Scaffold=Terminus-22026.06 | 24.3 | |
| GPT-5-MiniSize=-, Post-Training Data=-, Scaffold=Terminus-22026.06 | 24 | |
| Qwen 3 Coder 480BSize=480B, Post-Training Data=-, Scaffold=Terminus-22026.06 | 23.9 | |
| Grok 4Size=-, Post-Training Data=-, Scaffold=Terminus-22026.06 | 23.1 | |
| TerminalTraj-Qwen2.5-Coder-32BSize=32B, Post-Training Data=50.7K, Scaffold=Terminus-22026.06 | 22 | |
| LiberCoder-Qwen3-32BSize=32B, Post-Training Data=49.5K, Scaffold=OpenHands2026.06 | 19.5 | |
| GPT-OSS-120BSize=120B, Post-Training Data=-, Scaffold=Terminus-22026.06 | 18.7 | |
| TermiGen-Qwen3-32BSize=32B, Post-Training Data=3.5K, Scaffold=Bashagent2026.06 | 18 | |
| Nemotron-Terminal-Qwen3-8BSize=8B, Post-Training Data=490.5K, Scaffold=Terminus-22026.06 | 13 | |
| Terminal-Lego-Qwen3-8BSize=8B, Post-Training Data=15.3K, Scaffold=Terminus-22026.06 | 11.8 | |
| TerminalTraj-Qwen2.5-Coder-7BSize=7B, Post-Training Data=50.7K, Scaffold=Terminus-22026.06 | 10.1 | |
| Qwen3-32BSize=32B, Post-Training Data=-, Scaffold=Terminus-22026.06 | 3.4 | |
| GPT-OSS-20BSize=20B, Post-Training Data=-, Scaffold=Terminus-22026.06 | 3.1 | |
| Qwen3-8BSize=8B, Post-Training Data=-, Scaffold=Terminus-22026.06 | 2.5 |