Long-video Question Answering on TemporalBench
73.2Binary AccuracyGPT-4o
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o2025.01 | 73.2 | |
| LLaVA-Video-72B2025.01 | 72.4 | |
| Qwen2-VL-72B2025.01 | 70.2 | |
| Tarsier-34B2025.01 | 66.7 | |
| Gemini-1.5-Pro2025.01 | 66.4 | |
| Tarsier2-7BNumber of frames=128f2025.01 | 65.3 | |
| LLaVA-Video-7B2025.01 | 63.6 | |
| Qwen2-VL-7B2025.01 | 62 | |
| Tarsier-7B2025.01 | 56.9 |