OpenVLM Video - TempCompass - Captioning: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 36 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | InternVL3-78B | 74.9 |
| 2 | Qwen 2 VL 72B | 72.95 |
| 3 | GPT-4o (2024-08-06) | 69.21 |
| 4 | Gemini 2.0 Flash | 63.82 |
| 5 | Claude 3.5 Sonnet | 63.42 |
| 6 | InternVL3-8B | 61.08 |
| 7 | Gemini 1.5 Flash | 55.94 |
| 8 | Qwen 2 VL 7B | 55.34 |
| 9 | InternVL2-8B | 53.94 |
| 10 | Phi-4 Multimodal Instruct | 37.03 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-video-tempcompass-captioning · How It Works · Data refreshed daily, snapshot 2026-09-19.