Evaluation using Short, Medium, and Long video scenarios, across reasoning types: temporal, spatial, and intent.
| Difficulty | Model | Size | Over. Avg. | Short Avg. | Short Temp. | Short Spatial | Short Intent | Medium Avg. | Medium Temp. | Medium Spatial | Medium Intent | Long Avg. | Long Temp. | Long Spatial | Long Intent |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Hard | GPT 4o 🥇 | - | 21.26 | 24.05 | 25.32 | 30.34 | 16.5 | 25.19 | 27.57 | 31.07 | 16.91 | 14.66 | 5.5 | 30.5 | 8 |
| Gemini 1.5 pro | - | 20.55 | 23.2 | 23.88 | 23.2 | 22.5 | 21.61 | 26.66 | 19.04 | 19.12 | 16.84 | 6 | 25.5 | 19 | |
| InternVL2.5 | 26B | 20.55 | 20.33 | 25 | 28.5 | 8.5 | 25.66 | 31 | 32 | 14 | 15.66 | 13 | 17 | 17 | |
| InternVL2.5 | 8B | 20.45 | 19.34 | 19 | 30.5 | 8.5 | 24.66 | 31 | 30 | 13 | 17.34 | 10.5 | 31.5 | 10 | |
| InternVL2.5 | 4B | 17.45 | 17 | 16 | 19 | 15 | 21 | 25 | 21 | 17 | 14.34 | 11.5 | 26 | 5.5 | |
| LLaVA Next | 32B | 17.05 | 19.67 | 15 | 33 | 11 | 14 | 9 | 22 | 11 | 17.5 | 7.5 | 35 | 10 | |
| LLaVA Video | 7B | 17.28 | 18 | 13 | 31.5 | 9.5 | 18.67 | 16 | 26 | 14 | 15.16 | 7.5 | 29 | 9 | |
| LLaVA OneVision | 7B | 14.67 | 15.16 | 8.5 | 27.5 | 9.5 | 15.34 | 15 | 17 | 14 | 13.5 | 8 | 23.5 | 9 | |
| Qwen2.5 VL | 32B | 19.44 | 19.66 | 8.5 | 35 | 15.5 | 25.33 | 25 | 24 | 27 | 13.33 | 2 | 28 | 10 | |
| Qwen2.5 VL | 7B | 19.72 | 22.66 | 8.5 | 30 | 29.5 | 22.66 | 21 | 31 | 16 | 13.84 | 3.5 | 30 | 8 | |
| Qwen2.5 VL | 3B | 19.05 | 19.5 | 12.5 | 32.5 | 13.5 | 20 | 22 | 26 | 12 | 17.67 | 10 | 29.5 | 13.5 | |
| Medium | GPT 4o 🥇 | - | 37.72 | 42.08 | 43.24 | 55.5 | 27.5 | 31.95 | 39.84 | 30.34 | 25.66 | 39 | 44.5 | 37 | 35.5 |
| Gemini 1.5 pro | - | 36.34 | 38.73 | 37.21 | 45 | 34 | 35.09 | 33.66 | 47.11 | 24.5 | 35.17 | 21 | 53.5 | 31 | |
| InternVL2.5 | 26B | 31.89 | 33.66 | 33.5 | 54 | 13.5 | 30.67 | 31 | 43 | 18 | 31.34 | 27.5 | 42.5 | 24 | |
| InternVL2.5 | 8B | 34.49 | 33.66 | 31.5 | 57.5 | 12 | 35 | 37 | 48 | 20 | 34.83 | 33 | 44.5 | 27 | |
| InternVL2.5 | 4B | 33.05 | 34.5 | 33 | 48.5 | 22 | 33.34 | 37 | 41 | 22 | 31.33 | 25.5 | 43 | 25.5 | |
| LLaVA Next | 32B | 23.05 | 26 | 17 | 44.5 | 16.5 | 18 | 16 | 25 | 13 | 25.16 | 20.5 | 38 | 17 | |
| LLaVA Video | 7B | 24.84 | 25.16 | 22 | 35 | 21 | 24.34 | 26 | 27 | 20 | 25 | 14.5 | 42.5 | 18 | |
| LLaVA OneVision | 7B | 20.17 | 19.66 | 23 | 32 | 16 | 18.67 | 19 | 20 | 17 | 22.16 | 16 | 32.5 | 18 | |
| Qwen2.5 VL | 32B | 30.95 | 30.5 | 16.5 | 46 | 29 | 32 | 31 | 40 | 25 | 30.34 | 14 | 50 | 27 | |
| Qwen2.5 VL | 7B | 28.95 | 31.84 | 26.5 | 33 | 36 | 28.34 | 28 | 33 | 24 | 26.66 | 25.5 | 23 | 31.5 | |
| Qwen2.5 VL | 3B | 29.73 | 32.17 | 25.5 | 44 | 27 | 25.67 | 24 | 37 | 16 | 31.34 | 31 | 35.5 | 27.5 | |
| Easy | GPT 4o | - | 41.42 | 43.84 | 44.5 | 37.53 | 49.5 | 41.91 | 39.45 | 41.45 | 44.84 | 38.5 | 44.5 | 27.5 | 43.5 |
| Gemini 1.5 pro | - | 44.5 | 48.33 | 48 | 47 | 50 | 39.46 | 48.51 | 34.36 | 35.5 | 45.84 | 46.5 | 47 | 44 | |
| InternVL2.5 | 26B | 44.33 | 48.16 | 49 | 51.5 | 44 | 40 | 43 | 45 | 32 | 44.83 | 46 | 51 | 37.5 | |
| InternVL2.5 | 8B | 44.27 | 46.17 | 41.5 | 53 | 44 | 40 | 45 | 42 | 33 | 46.66 | 57 | 52 | 31 | |
| InternVL2.5 | 4B | 42.61 | 48.33 | 44 | 55 | 46 | 38.33 | 39 | 41 | 35 | 41.16 | 39.5 | 54 | 30 | |
| LLaVA Next | 32B | 32.23 | 37.34 | 35.5 | 43.5 | 33 | 26.33 | 24 | 23 | 32 | 33.17 | 27.5 | 40 | 32 | |
| LLaVA Video | 7B | 32.33 | 33.16 | 32 | 34.5 | 33 | 34 | 36 | 37 | 29 | 29.84 | 25.5 | 31 | 33 | |
| LLaVA OneVision | 7B | 31.5 | 32.66 | 32.5 | 35.5 | 30 | 29.34 | 30 | 34 | 24 | 32.5 | 31.5 | 33 | 33 | |
| Qwen2.5 VL 🥇 | 32B | 47.84 | 50.5 | 46 | 53 | 52.5 | 46 | 43 | 46 | 49 | 47 | 43.5 | 52 | 45.5 | |
| Qwen2.5 VL | 7B | 40.28 | 42.33 | 41.5 | 30 | 55.5 | 37 | 40 | 29 | 42 | 41.5 | 44.5 | 29 | 51 | |
| Qwen2.5 VL | 3B | 42.84 | 46.33 | 37 | 53.5 | 48.5 | 38.67 | 40 | 44 | 32 | 43.5 | 38 | 52 | 40.5 |