Audio-Visual Understanding and Reasoning on WorldSense
48AVGGemini-1.5-Pro
Evaluation results
| Method | Links | |
|---|---|---|
| Gemini-1.5-Pro2026.06 | 48 | |
| OmniAgentSize=7B2026.06 | 47.2 | |
| Qwen2.5-OmniSize=7B2026.06 | 45.4 | |
| GPT-4o2026.06 | 42.6 | |
| Unified-IO-2Size=8B2026.06 | 25.9 | |
| VideoLLaMA2Size=7B2026.06 | 25.4 |