OpenVLM Video - TempCompass: leaderboard
OpenVLM Video leaderboard slice for TempCompass temporal video understanding tasks.
Metric: Normalized Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 37 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2 VL 72B | 79.36 | #460 |
| 2 | InternVL3-38B | 78.49 | #405 |
| 3 | InternVL3-78B | 76.98 | #352 |
| 4 | Qwen 2.5 VL 7B | 72.04 | #643 |
| 5 | Qwen2.5-Omni-7B | 70.72 | #624 |
| 6 | InternVL3-8B | 70.42 | #620 |
| 7 | Qwen 2 VL 7B | 69.56 | #750 |
| 8 | Claude 3.5 Sonnet | 68.29 | #350 |
| 9 | InternVL2-8B | 65.42 | #817 |
| 10 | Gemini 1.5 Flash | 63.63 | #618 |
| 11 | MiniCPM-V-2.6 | 59.64 | #836 |
| 12 | Gemini 2.0 Flash | 57.85 | #342 |
| 13 | InternVL2.5-78B | 55.95 | #431 |
| 14 | Phi-4 Multimodal Instruct | 50.4 | #906 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-video-tempcompass · How It Works · Data refreshed daily, snapshot 2026-10-20.