OpenVLM Video Leaderboard: leaderboard
OpenVLM Video Leaderboard evaluates video-capable vision-language models on video benchmarks including Video-MME, MVBench, MMBench-Video, TempCompass, and MLVU. The pinned metric is Avg Score, measuring the aggregated score across the mirrored source leaderboard tables. Source: BenchmarkList’s attributed result table.
Metric: Avg Score. Source: benchmarklist.com. 37 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-78B | 72.7 | #361 |
| 2 | InternVL3-38B | 71.9 | #413 |
| 3 | InternVL2.5-78B | 68.1 | #434 |
| 4 | Qwen 2 VL 72B | 67.7 | #450 |
| 5 | InternVL3-8B | 66.8 | #624 |
| 6 | Gemini 2.0 Flash | 64.6 | #351 |
| 7 | Qwen2.5-Omni-7B | 64.5 | #633 |
| 8 | GPT-4o (2024-08-06) | 63.1 | #343 |
| 9 | Qwen 2.5 VL 7B | 62.8 | #641 |
| 10 | Qwen 2 VL 7B | 61.2 | #743 |
| 11 | Gemini 1.5 Flash | 57 | #620 |
| 12 | InternVL2-8B | 56.4 | #813 |
| 13 | MiniCPM-V-2.6 | 54.7 | #827 |
| 14 | Claude 3.5 Sonnet | 45.9 | #362 |
| 15 | Phi-4 Multimodal Instruct | 42.3 | #922 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-video-leaderboard · How It Works · Data refreshed daily, snapshot 2026-11-05.