Multi-Turn Tool Calling on BFCL Multi-Turn
55.5Base PerformanceTool-N1
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Tool-N12026.04 | 55.5 | 45.5 | 38 | 38.5 | 44.4 | |
| TRUSTEE2026.04 | 54 | 49 | 40 | 42.5 | 46.4 | |
| EnvScaler2026.04 | 53.5 | 45.5 | 39 | 40 | 44.5 | |
| ToolRL2026.04 | 52 | 45 | 35 | 33 | 41.3 | |
| QWEN3-8B2026.04 | 49.5 | 44 | 34 | 32 | 39.9 | |
| AWM2026.04 | 48.5 | 44 | 36 | 30.5 | 39.8 | |
| BaseModel=Granite-4.1-8B2026.06 | 44.5 | 23.5 | 26 | 39.5 | 33.4 | |
| PROVEModel=Granite-4.1-8B2026.06 | 43 | 30 | 24 | 37 | 33.5 | |
| PROVEModel=Qwen3-4B2026.06 | 35.5 | 29 | 18.5 | 33 | 29 | |
| PROVEModel=Qwen3-8B2026.06 | 34 | 32 | 21.5 | 20 | 26.9 | |
| BaseModel=Qwen3-8B2026.06 | 30.5 | 31.5 | 19.5 | 22.5 | 26 | |
| BaseModel=Qwen3-4B2026.06 | 23.5 | 14 | 14.5 | 23 | 18.8 | |
| PROVEModel=Qwen2.5-7B2026.06 | 22.5 | 18 | 13.5 | 13 | 16.8 | |
| BaseModel=Qwen2.5-7B2026.06 | 17 | 14.5 | 10 | 11 | 13.1 | |
| Simia2026.04 | 2.5 | 2 | 1.5 | 2 | 2 |