Tool Use on τ²-Bench (out-of-distribution)
54.9Retail ScorexLAM2-70B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| xLAM2-70BSize=70B2026.02 | 54.9 | 46 | 29.3 | 43.4 | |
| D-CORE-14BSize=14B2026.02 | 53.5 | 44.1 | 34.9 | 44.2 | |
| xLAM2-32BSize=32B2026.02 | 53.1 | 41.6 | 26.1 | 40.3 | |
| xLAM2-8BSize=8B2026.02 | 51.3 | 35.4 | 22.4 | 36.4 | |
| D-CORE-8BSize=8B2026.02 | 49.8 | 38.4 | 30.5 | 39.6 | |
| Qwen3-32BSize=32B2026.02 | 49.1 | 36 | 28.4 | 37.8 | |
| Qwen3-14BSize=14B2026.02 | 46.5 | 30 | 31.7 | 36.1 | |
| Qwen3-8BSize=8B2026.02 | 41.5 | 31.3 | 26.3 | 33 |