Audio-Visual Question on WorldSense (Robustness Evaluation)
59.7Accuracy (Clean)Gemini-3.1-Flash-Lite-preview
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Gemini-3.1-Flash-Lite-previewModel=Gemini-3.1-Flash-Lite-preview2026.04 | 59.7 | 36.21 | 14.58 | 48.33 | |
| Qwen3-Omni-30BModel=Qwen3-Omni-30B2026.04 | 55.72 | 24.87 | 14.35 | 61.39 | |
| Qwen2.5-Omni-7BModel=Qwen2.5-Omni-7B2026.04 | 49.9 | 21.07 | 16.59 | 64.03 | |
| Gemini-2.5-Flash-LiteModel=Gemini-2.5-Flash-Lite2026.04 | 49.33 | 29.08 | 19.66 | 56.27 | |
| ChatBridgeModel=ChatBridge2026.04 | 33.57 | 31.36 | 27.42 | 29.82 | |
| PandaGPTModel=PandaGPT2026.04 | 29.48 | 29.4 | 25.27 | 25.75 |