Tool Use on τ-Bench (TauB) V2 (accuracy)
91.6AccuracyQwen3.5-122B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3.5-122BDeployment=Local model2026.05 | 91.6 | |
| Best LocalConfig=Oracle local routing frontier2026.05 | 91.6 | |
| Gemma4-26BDeployment=Local model2026.05 | 91.3 | |
| Gemini 3.1 ProDeployment=Cloud baseline2026.05 | 90.8 | |
| Qwen3.5-35BDeployment=Local model2026.05 | 90.2 | |
| Claude Opus 4.6Deployment=Cloud baseline2026.05 | 89.5 | |
| GPT 5.4Deployment=Cloud baseline2026.05 | 89.2 | |
| Qwen3.5-27BDeployment=Local model2026.05 | 88.4 | |
| Qwen3.5-9BDeployment=Local model2026.05 | 77.1 | |
| Gemma4-E4BDeployment=Local model2026.05 | 56.1 | |
| Nemotron-Super-120BDeployment=Local model2026.05 | 36.8 | |
| Granite 4.0 H-SmallDeployment=Local model2026.05 | 17.5 | |
| Granite 3.3 8BDeployment=Local model2026.05 | 10.5 |