Multimodal Agent Performance on CV-Bench
83.7AccuracyToolGate
Evaluation Results
| Method | Links | |
|---|---|---|
| ToolGateBackbone=235B-FP8, Training regime=Cross-domain2026.06 | 83.7 | |
| BaselineBackbone=235B-FP8, Training regime=None2026.06 | 82.7 | |
| ToolGateBackbone=30B, Training regime=In-domain2026.06 | 80 | |
| BaselineBackbone=30B, Training regime=None2026.06 | 78.8 | |
| ToolGateBackbone=30B, Training regime=Cross-domain2026.06 | 77.6 |