Tool Use on StableToolBench Overall Average
70.3SL (Success Rate)Trace-Based
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Trace-BasedEvaluation Protocol=Trace-based2026.02 | 70.3 | 54.6 | |
| Play2PromptEvaluation Protocol=Trace-based2026.02 | 69.8 | 52.5 | |
| D2Evaluation Protocol=Trace-based2026.02 | 69.5 | 52.5 | |
| DRAFTEvaluation Protocol=Trace-based2026.02 | 68.1 | 50 | |
| D0Evaluation Protocol=Trace-based2026.02 | 67.3 | 48 | |
| D1Evaluation Protocol=Trace-based2026.02 | 66.5 | 49.4 |