Audio-Visual Video Understanding on JointAVBench truncated <= 5min
66.37Overall ScoreOurs(Qwen3-Omni)
Evaluation Results
| Method | Links | |
|---|---|---|
| Ours(Qwen3-Omni)Size=30B2026.06 | 66.37 | |
| Qwen3-OmniSize=30B2026.06 | 63.17 | |
| Ours(Qwen2.5-Omni)Size=7B2026.06 | 60.75 | |
| OmniVinciSize=9B2026.06 | 57.55 | |
| Uni-MoE-2.0-OmniSize=30B2026.06 | 57.55 | |
| Ours(VITA-1.5)Size=8B2026.06 | 57.41 | |
| video-SALMONN 2+Size=7B2026.06 | 56.39 | |
| MiniCPM-o 4.5Size=9B2026.06 | 55.18 | |
| Qwen2.5-OmniSize=7B2026.06 | 54.44 | |
| VITA-1.5Size=8B2026.06 | 44.77 |