| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| ToolBench | ExpRAG | API Success Rate88 | 28 | 2mo ago | |
| Ours (test) | Gemini 2.5-Flash | Solve Precision (P)52.78 | 27 | 3mo ago | |
| Tool use | GRPO | Mean Accuracy @1661.31 | 24 | 1mo ago | |
| ToolQA | Accuracy47.97 | 20 | 1mo ago | ||
| ToolAlpaca | MOPD | Accuracy66.73 | 20 | 2mo ago | |
| ShopTrajQA 32k context | Accuracy62.6 | 14 | 1mo ago | ||
| τ-Bench | Avg Accuracy63.9 | 14 | 4mo ago | ||
| ShopTrajQA 64k context | Qwen3-4B-SFT-RLVR | Accuracy60.1 | 13 | 1mo ago | |
| BFCL V4 | Llama-3.1-8B-Instruct + RODS (ours) | Average Performance18 | 6 | 1mo ago | |
| Tau-Telecom | NanoV3 Elastic-30B | Success Rate33.33 | 5 | 2mo ago | |
| Tau-Retail | NanoV3 Elastic-30B | Success Rate59.36 | 5 | 2mo ago | |
| Tau-Airline | Qwen3-30B-A3B | Success Rate52.67 | 5 | 2mo ago |