Multi-tool calling on SEAL-Tools
86.5TSAJTPRO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| JTPROBackbone=GPT-52026.04 | 86.5 | 65.2 | 33.6 | |
| GEPABackbone=GPT-52026.04 | 85.2 | 61.3 | 31.1 | |
| BaseBackbone=GPT-52026.04 | 84.5 | 56.3 | 28.8 | |
| JTPROBackbone=o3-mini2026.04 | 83.9 | 60.1 | 30.1 | |
| JTPROBackbone=GPT-4o2026.04 | 82.3 | 53.51 | 27.5 | |
| BaseBackbone=o3-mini2026.04 | 82.2 | 52.3 | 26.3 | |
| GEPABackbone=GPT-4o2026.04 | 82.11 | 43.51 | 24.51 | |
| GEPABackbone=o3-mini2026.04 | 82.1 | 56.9 | 27.5 | |
| BaseBackbone=GPT-4o2026.04 | 81 | 40.4 | 23 |