API Tool Use on Real-World APIs (test)
98.3TCRNaviAgent
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| NaviAgentModel=DeepSeek-V32025.06 | 98.3 | 63.3 | 5.11 | 35 | |
| NaviAgentModel=Qwen2.5-32B2025.06 | 86.7 | 53.3 | 4.84 | 27 | |
| α-UMIModel=DeepSeek-V32025.06 | 85 | 48.3 | 6.17 | 39 | |
| α-UMIModel=Qwen2.5-32B2025.06 | 78.3 | 41.6 | 6.04 | 32 | |
| α-UMIModel=Qwen2.5-14B2025.06 | 76.7 | 31.7 | 5.8 | 27 | |
| NaviAgentModel=Qwen2.5-14B2025.06 | 65 | 36.7 | 4.9 | 25 | |
| ReActModel=DeepSeek-V32025.06 | 55 | 33.3 | 3.75 | 23 | |
| ToolLLMModel=Qwen2.5-14B2025.06 | 53.3 | 23.3 | 4.1 | 19 | |
| ToolLLMModel=DeepSeek-V32025.06 | 53.3 | 35 | 4.05 | 25 | |
| ToolLLMModel=Qwen2.5-32B2025.06 | 48.3 | 30 | 4 | 23 | |
| ReActModel=Qwen2.5-32B2025.06 | 35 | 25 | 3.8 | 19 | |
| ReActModel=Qwen2.5-14B2025.06 | 31.7 | 21.7 | 3.7 | 16 |