Multimodal Reasoning on OmniBench
54.9Causal ScoreBase
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| BaseModel=Qwen3-Omni-30B-A3B-Instruct2026.05 | 54.9 | 40.5 | 31 | 48.8 | |
| ReGapModel=Qwen3-Omni-30B-A3B-Instruct2026.05 | 54.5 | 41.1 | 44.8 | 49.6 | |
| ReGapModel=MiniCPM-o-4.52026.05 | 50.6 | 50.3 | 27.6 | 49.2 | |
| BaseModel=MiniCPM-o-4.52026.05 | 50 | 47.2 | 27.6 | 47.8 | |
| Refusal SteeringModel=Qwen3-Omni-30B-A3B-Instruct2026.05 | 40.6 | 26.4 | 34.5 | 35.6 | |
| Refusal SteeringModel=MiniCPM-o-4.52026.05 | 32.1 | 28.8 | 20.7 | 30.4 | |
| ReGapModel=Qwen2.5-Omni-3B2026.05 | 31.8 | 42.9 | 20.7 | 34.8 | |
| BaseModel=Qwen2.5-Omni-3B2026.05 | 28.6 | 42.3 | 17.2 | 32.4 | |
| Refusal SteeringModel=Qwen2.5-Omni-3B2026.05 | 13.6 | 19.6 | 3.5 | 15 |