Tool-use API Generalization on ToolBench G2
78.2Pass RateEVOTOOL
Evaluation Results
| Method | Links | |
|---|---|---|
| EVOTOOLBackbone=GPT-4o2026.03 | 78.2 | |
| DRAFTBackbone=GPT-4o2026.03 | 76.8 | |
| EASYTOOLBackbone=GPT-4o2026.03 | 75.1 | |
| EvoPromptBackbone=GPT-4o2026.03 | 69.8 | |
| ANYTOOLBackbone=GPT-4o2026.03 | 68.8 | |
| OPROBackbone=GPT-4o2026.03 | 67.3 | |
| PromptBreederBackbone=GPT-4o2026.03 | 66.5 | |
| EVOTOOLBackbone=Qwen3-8B2026.03 | 66.5 | |
| DRAFTBackbone=Qwen3-8B2026.03 | 65.3 | |
| ReActBackbone=GPT-4o2026.03 | 64.7 | |
| EASYTOOLBackbone=Qwen3-8B2026.03 | 61.8 | |
| EvoPromptBackbone=Qwen3-8B2026.03 | 59.3 | |
| Plan-and-SolveBackbone=GPT-4o2026.03 | 59.2 | |
| AdaPlannerBackbone=GPT-4o2026.03 | 58.5 | |
| ANYTOOLBackbone=Qwen3-8B2026.03 | 58.5 | |
| OPROBackbone=Qwen3-8B2026.03 | 57.2 | |
| PromptBreederBackbone=Qwen3-8B2026.03 | 56.5 | |
| ReActBackbone=Qwen3-8B2026.03 | 55 | |
| CoTBackbone=GPT-4o2026.03 | 50.8 | |
| Plan-and-SolveBackbone=Qwen3-8B2026.03 | 50.3 | |
| AdaPlannerBackbone=Qwen3-8B2026.03 | 49.7 | |
| CoTBackbone=Qwen3-8B2026.03 | 43.2 |