Long-video understanding on VideoMME long
69AccuracyGemini-2.0-Flash
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Gemini-2.0-FlashAugmentation=Light-Omni2026.07 | 69 | — | — | 59.6 | |
| Light-OmniBackbone=Qwen2.5-Omni-7B2026.07 | 66.1 | 2.221 | 24.03 | 58 | |
| GPT-4o2026.07 | 65.3 | — | — | 48.1 | |
| Qwen3-VL-8BAugmentation=Light-Omni2026.07 | 65.1 | 7.02 | 46.3 | 57.2 | |
| Qwen2.5-VL-72B2026.07 | 63.4 | — | — | 56.1 | |
| Gemini-2.0-Flash2026.07 | 63 | — | — | 55.8 | |
| M3-Agent2026.07 | 61.8 | 25.5 | 62.4 | 55.6 | |
| Qwen3-VL-8B2026.07 | 61 | 14.9 | 29.9 | 54.7 | |
| RAG-Rewrite2026.07 | 60.8 | 8.5 | 27.7 | 54.2 | |
| Qwen2.5-VL-7BAugmentation=Light-Omni2026.07 | 59.6 | 8.56 | 46.72 | 54.5 | |
| Naive RAG2026.07 | 58.4 | 7.3 | 27.6 | 52.7 | |
| Qwen2.5-VL-7B2026.07 | 55.4 | 51.7 | 113.2 | 49.6 | |
| Qwen2.5-Omni-7B2026.07 | 55.3 | 48.8 | 81.7 | 48.5 | |
| HippoMM2026.07 | 41.6 | 16.1 | 36.3 | 39.9 | |
| MovieChat2026.07 | 19.4 | — | 30.7 | 21 |