OpenVLM Video - TempCompass - Multi-choice: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 35 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2 VL 72B | 77.85 |
| 2 | InternVL3-78B | 75.95 |
| 3 | InternVL2.5-78B | 73.1 |
| 4 | Qwen 2 VL 7B | 70.44 |
| 5 | InternVL3-8B | 70.38 |
| 6 | GPT-4o (2024-08-06) | 69.87 |
| 7 | Claude 3.5 Sonnet | 66.71 |
| 8 | InternVL2-8B | 64.81 |
| 9 | Gemini 2.0 Flash | 64.43 |
| 10 | Gemini 1.5 Flash | 63.16 |
| 11 | Phi-4 Multimodal Instruct | 48.29 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-video-tempcompass-multi-choice · How It Works · Data refreshed daily, snapshot 2026-09-19.