Temporal Video Understanding on MotionBench
65.4AccuracyGPT-4o
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o2026.06 | 65.4 | |
| Gemini 1.5 ProSource=[59]2026.06 | 62.6 | |
| Molmo2-8B2026.06 | 62.2 | |
| Gemini 1.5 ProSource=[60]2026.06 | 62 | |
| PLM-7B2026.06 | 61.4 | |
| VideoChat-Flash-7B2026.06 | 60.6 | |
| Molmo2-O-7B2026.06 | 60.6 | |
| GPT-4o mini2026.06 | 59.9 | |
| AdaCodecToken budget=comparable, Backbone=Qwen2-VL-7B2026.06 | 59.9 | |
| MiniCPM-V-2.6-8B2026.06 | 59.7 | |
| Gemini 1.5 Flash2026.06 | 59.3 | |
| GLM-4V-9B2026.06 | 59 | |
| AdaCodecToken budget=1/7, Backbone=Qwen2-VL-7B2026.06 | 58.8 | |
| Claude 3.5 Sonnet2026.06 | 58.5 | |
| Qwen2-VL-7B2026.06 | 56.9 | |
| InternVL2-8B2026.06 | 56.6 | |
| Eagle-X5-7B2026.06 | 55.7 | |
| Key-VL-1.5-8B2026.06 | 55.1 | |
| LLaVA-Video-7B2026.06 | 54.2 |