AutoEval-Video — leaderboard
AutoEval-Video evaluates large vision-language models on open-ended video question answering across nine video perception and reasoning dimensions.
Metric: Overall Accuracy (%). Source: huggingface.co. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4V | 22.2 |
| 2 | VideoChat | 13.4 |
| 3 | Video-LLaMA | 11.2 |
| 4 | LLaVA-1.5 | 8.5 |
| 5 | InstructBLIP | 7.9 |
| 6 | Qwen-VL | 7 |
| 7 | Video-ChatGPT | 6.7 |
| 8 | BLIP-2 | 0.3 |
Interactive version: theaggregate.ai/benchmark?slug=autoeval-video · How the rankings work · Data refreshed daily, snapshot 2026-07-22.