Tool-use API Generalization on ToolBench G1 v1
83.5Pass RateEVOTOOL
Evaluation Results
| Method | Links | |
|---|---|---|
| EVOTOOLBackbone=GPT-4o, Optimization Strategy=Ours2026.03 | 83.5 | |
| DRAFTBackbone=GPT-4o, Optimization Strategy=Single-aspect optimization2026.03 | 82.1 | |
| EASYTOOLBackbone=GPT-4o, Optimization Strategy=Single-aspect optimization2026.03 | 80.3 | |
| EVOTOOLBackbone=Qwen3-8B, Optimization Strategy=Ours2026.03 | 73.5 | |
| ANYTOOLBackbone=GPT-4o, Optimization Strategy=Single-aspect optimization2026.03 | 73.3 | |
| DRAFTBackbone=Qwen3-8B, Optimization Strategy=Single-aspect optimization2026.03 | 72.2 | |
| EASYTOOLBackbone=Qwen3-8B, Optimization Strategy=Single-aspect optimization2026.03 | 70.7 | |
| EvoPromptBackbone=GPT-4o, Optimization Strategy=Monolithic optimization2026.03 | 70.1 | |
| OPROBackbone=GPT-4o, Optimization Strategy=Monolithic optimization2026.03 | 69.3 | |
| PromptBreederBackbone=GPT-4o, Optimization Strategy=Monolithic optimization2026.03 | 68.5 | |
| ReActBackbone=GPT-4o, Optimization Strategy=Hand-crafted policies2026.03 | 68.2 | |
| ANYTOOLBackbone=Qwen3-8B, Optimization Strategy=Single-aspect optimization2026.03 | 64.5 | |
| Plan-and-SolveBackbone=GPT-4o, Optimization Strategy=Hand-crafted policies2026.03 | 63.3 | |
| AdaPlannerBackbone=GPT-4o, Optimization Strategy=Single-aspect optimization2026.03 | 62.2 | |
| EvoPromptBackbone=Qwen3-8B, Optimization Strategy=Monolithic optimization2026.03 | 61.7 | |
| OPROBackbone=Qwen3-8B, Optimization Strategy=Monolithic optimization2026.03 | 61 | |
| PromptBreederBackbone=Qwen3-8B, Optimization Strategy=Monolithic optimization2026.03 | 60.3 | |
| ReActBackbone=Qwen3-8B, Optimization Strategy=Hand-crafted policies2026.03 | 60 | |
| Plan-and-SolveBackbone=Qwen3-8B, Optimization Strategy=Hand-crafted policies2026.03 | 55.7 | |
| AdaPlannerBackbone=Qwen3-8B, Optimization Strategy=Single-aspect optimization2026.03 | 54.7 | |
| CoTBackbone=GPT-4o, Optimization Strategy=Hand-crafted policies2026.03 | 53.5 | |
| CoTBackbone=Qwen3-8B, Optimization Strategy=Hand-crafted policies2026.03 | 47.1 |