VideoInstruct (gpt-score): leaderboard
VideoInstruct benchmark (Video-ChatGPT) — judge-scored video instruction following; overall judge score.
Metric: gpt-score. Source: huggingface.co. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | PPLLaVA-7B | 4.21 | |
| 2 | PLLaVA-34B | 3.9 | |
| 3 | TS-LLaVA-34B | 3.86 | |
| 4 | SlowFast-LLaVA-34B | 3.84 | |
| 5 | ST-LLM | 3.74 | |
| 6 | VideoGPT+ | 3.74 | |
| 7 | VideoChat2_HD_mistral | 3.64 | |
| 8 | MiniGPT4-video-7B | 3.57 | |
| 9 | VideoChat2 | 3.51 | |
| 10 | Chat-UniVi | 3.46 |
Interactive version: theaggregate.ai/benchmark?slug=videoinstruct-gpt-score · How It Works · Data refreshed daily, snapshot 2026-10-20.