PinchBench
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Agent Task | PinchBench (PB) | Accuracy100 | 21 | |
| Long-horizon task solving | PINCHBENCH 23 tasks original | Best Success Rate93.3 | 19 | |
| Agent task completion | PinchBench | Pass@188.7 | 17 | |
| Agentic reasoning | PinchBench | Score91.3 | 16 | |
| Long-horizon task solving | PINCHBENCH 123 tasks expanded | Best Success Rate90.3 | 15 | |
| Real-World Agent | PinchBench | Average Score82.3 | 15 | |
| Agent Utility Evaluation | PinchBench v2.0.0 | Utility Score89.6 | 12 | |
| Context Management and LLM Evaluation | PinchBench Continuous Mode | Overall Score81.3 | 11 | |
| Context Management and LLM Evaluation | PinchBench Isolated Mode | Overall Score81 | 11 | |
| Agentic Evaluation | PinchBench | Avg@387.6 | 6 | |
| Task Performance Evaluation | PinchBench v2.0.0 | Best Score89.31 | 6 | |
| Task Performance Evaluation | PinchBench v1.2.0 | Best Score90.1 | 6 | |
| Agent & OpenClaw | PinchBench | Accuracy83.7 | 5 |
Showing 13 of 13 rows