Tool-use API Generalization on ToolBench G3
71.5Pass RateEVOTOOL
Evaluation Results
| Method | Links | |
|---|---|---|
| EVOTOOLBackbone=GPT-4o2026.03 | 71.5 | |
| DRAFTBackbone=GPT-4o2026.03 | 68.4 | |
| EASYTOOLBackbone=GPT-4o2026.03 | 66.2 | |
| ANYTOOLBackbone=GPT-4o2026.03 | 61 | |
| EvoPromptBackbone=GPT-4o2026.03 | 59.2 | |
| OPROBackbone=GPT-4o2026.03 | 58.9 | |
| EVOTOOLBackbone=Qwen3-8B2026.03 | 58.6 | |
| ReActBackbone=GPT-4o2026.03 | 57.9 | |
| DRAFTBackbone=Qwen3-8B2026.03 | 56.1 | |
| Plan-and-SolveBackbone=GPT-4o2026.03 | 55.3 | |
| PromptBreederBackbone=GPT-4o2026.03 | 54.7 | |
| EASYTOOLBackbone=Qwen3-8B2026.03 | 54.3 | |
| ANYTOOLBackbone=Qwen3-8B2026.03 | 50 | |
| AdaPlannerBackbone=GPT-4o2026.03 | 48.8 | |
| EvoPromptBackbone=Qwen3-8B2026.03 | 48.5 | |
| CoTBackbone=GPT-4o2026.03 | 48.4 | |
| OPROBackbone=Qwen3-8B2026.03 | 48.3 | |
| ReActBackbone=Qwen3-8B2026.03 | 47.5 | |
| Plan-and-SolveBackbone=Qwen3-8B2026.03 | 45.3 | |
| PromptBreederBackbone=Qwen3-8B2026.03 | 44.9 | |
| AdaPlannerBackbone=Qwen3-8B2026.03 | 40 | |
| CoTBackbone=Qwen3-8B2026.03 | 39.7 |