OpenVLM Video - TempCompass - Caption Matching: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 35 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2 VL 72B | 86.36 |
| 2 | InternVL3-78B | 84.5 |
| 3 | InternVL2.5-78B | 83.23 |
| 4 | GPT-4o (2024-08-06) | 81.44 |
| 5 | Qwen 2 VL 7B | 80.37 |
| 6 | InternVL3-8B | 79.04 |
| 7 | Claude 3.5 Sonnet | 78.71 |
| 8 | InternVL2-8B | 76.38 |
| 9 | Gemini 1.5 Flash | 70.66 |
| 10 | Phi-4 Multimodal Instruct | 62.28 |
| 11 | Gemini 2.0 Flash | 16.03 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-video-tempcompass-caption-matching · How It Works · Data refreshed daily, snapshot 2026-09-19.