Real-World Agent on PinchBench
82.3Average ScoreClaude Opus 4.6
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Claude Opus 4.62026.03 | 82.3 | 87.4 | |
| KAT-Coder-V22026.03 | 81.9 | 88.7 | |
| MiniMax M2.72026.03 | 81.8 | 87.1 | |
| GPT-5.42026.03 | 81.6 | 90.5 | |
| GLM-52026.03 | 80.3 | 86.4 | |
| SkillAdaptorBackbone=GPT-5.22026.05 | 76.6 | — | |
| Gemini 3.1 Pro2026.03 | 75.9 | 86.7 | |
| OpenSpaceBackbone=GPT-5.22026.05 | 75.8 | — | |
| Base ModelBackbone=GPT-5.22026.05 | 74.8 | — | |
| SkillAdaptorBackbone=GLM-52026.05 | 73.8 | — | |
| OpenSpaceBackbone=GLM-52026.05 | 72.3 | — | |
| Base ModelBackbone=GLM-52026.05 | 70.2 | — | |
| SkillAdaptorBackbone=Kimi-K2.52026.05 | 67.2 | — | |
| OpenSpaceBackbone=Kimi-K2.52026.05 | 66 | — | |
| Base ModelBackbone=Kimi-K2.52026.05 | 63.6 | — |