Video Question Answering on Video-MME (p@1, p@4, m@4)
67.7P@1GPT-5.2
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-5.2frames=642026.07 | 67.7 | 72.9 | 67.9 | |
| Qwen3-VL-4B-Instructparameters=4B2026.07 | 63.7 | 68.6 | 64.3 | |
| Qwen3-VL-8B-Instructparameters=8B2026.07 | 63.7 | 68.6 | 64.3 | |
| VSeek-VETLreward_type=VETL2026.07 | 61.4 | 72 | 62.8 | |
| VSeek-GPTbase_model=GPT-5.22026.07 | 61.2 | 77.8 | 65.3 | |
| Qwen3-VL-8B-Thinkingparameters=8B2026.07 | 60.3 | 73.4 | 63 | |
| VSeek-EMreward_type=exact-match2026.07 | 59.3 | 71.2 | 60.1 | |
| Qwen3-VL-4B-Thinkingparameters=4B2026.07 | 58.1 | 65.2 | 60 | |
| VideoTree2026.07 | 56.3 | 58 | 56.6 | |
| Naive RAG2026.07 | 55 | 63.3 | 56.5 | |
| InternVL-3.5-4Bparameters=4B2026.07 | 53.1 | 76.1 | 54.7 |