Audiovisual Video Captioning on SALMONN 2 (test)
10Miss Ratevideo-SALMONN2-7B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| video-SALMONN2-7BModality=A + V2025.10 | 10 | 12.9 | — | |
| GPT-4oModality=V2025.10 | 17 | 14.2 | — | |
| Omni-Captioner-7BModality=A + V2025.10 | 17.8 | 10.9 | — | |
| Gemini-2.5-ProSize=-2026.01 | 18.1 | 13.3 | 31.3 | |
| Gemini-2.5-ProModality=A + V2026.02 | 18.1 | 13.3 | 31.3 | |
| Gemini-3-ProModality=A + V2026.02 | 19.1 | 16.6 | 35.7 | |
| Gemini-2.5-FlashSize=-2026.01 | 19.3 | 13.9 | 33.3 | |
| Gemini-2.5-FlashModality=A + V2026.02 | 19.3 | 13.9 | 33.3 | |
| ASID-CaptionerSize=7B, Modality=A + V2026.02 | 20.5 | 15.4 | 35.9 | |
| AVoCaDOSize=7B2026.01 | 21.1 | 16.2 | 37.3 | |
| DiaDemSize=7B2026.01 | 21.1 | 15.5 | 36.5 | |
| AVoCaDOSize=7B, Modality=A + V2026.02 | 21.1 | 16.2 | 37.3 | |
| video-SALMONN-2Size=7B2026.01 | 21.2 | 17.6 | 38.8 | |
| video-SALMONN-2Size=7B, Modality=A + V2026.02 | 21.2 | 17.6 | 38.8 | |
| Gemini-3-ProSize=-2026.01 | 21.4 | 14.2 | 35.6 | |
| Gemini 1.5 ProModality=A + V2025.10 | 21.8 | 16.5 | — | |
| Qwen2.5-VL-7BModality=V2025.10 | 21.9 | 17.4 | — | |
| LLaVA-OneVision-7BModality=V2025.10 | 23.3 | 27.4 | — | |
| ASID-CaptionerSize=3B, Modality=A + V2026.02 | 23.4 | 18.3 | 41.7 | |
| Qwen2.5-Omni-7BModality=A + V2025.10 | 26.3 | 21.7 | — | |
| InternVideo2.5-7BModality=V2025.10 | 30.8 | 15 | — | |
| Qwen3-Omni-CaptionerSize=30B-A3B2026.01 | 31 | 16.6 | 47.6 | |
| Qwen3-Omni-CaptionerSize=30B-A3B, Modality=A + V2026.02 | 31 | 16.6 | 47.6 | |
| UGC-VideoCaptionerSize=3B2026.01 | 31.6 | 17 | 48.6 | |
| UGC-VideoCaptionerSize=3B, Modality=A + V2026.02 | 31.6 | 17 | 48.6 | |
| Qwen3-Omni-InstructSize=30B-A3B2026.01 | 32 | 13.6 | 45.6 | |
| Qwen3-Omni-InstructSize=30B-A3B, Modality=A + V2026.02 | 32 | 13.6 | 45.6 | |
| Qwen2.5-VLSize=7B, Modality=V2026.02 | 40.5 | 17 | 57.5 | |
| Qwen2.5-OmniSize=7B2026.01 | 41.7 | 15.4 | 57.1 | |
| Qwen2.5-OmniSize=7B, Modality=A + V2026.02 | 41.7 | 15.4 | 57.1 | |
| MiniCPM-o-2.6Size=8B, Modality=A + V2026.02 | 42.2 | 14.3 | 56.5 | |
| VideoLLaMA3-7BModality=V2025.10 | 44.9 | 11.6 | — | |
| ARC-Hunyuan-VideoSize=7B, Modality=A + V2026.02 | 45.7 | 12.5 | 58.2 | |
| HumanOmniV2Size=7B, Modality=A + V2026.02 | 49.2 | 12.3 | 61.6 | |
| video-SALMONN-13BModality=A + V2025.10 | 52.1 | 26.6 | — | |
| InternVL3.5Size=8B, Modality=V2026.02 | 53.8 | 25.5 | 79.4 | |
| VideoLLaMA2-7BModality=A + V2025.10 | 56.8 | 8.9 | — |