OpenVLM MME: leaderboard
OpenCompass OpenVLM evaluation of MME: perception and cognition tests for multimodal models covering OCR, object existence, position, color, celebrity, artwork, commonsense, numerical calculation, and text translation.
Metric: Overall Score. Source: arxiv.org. Source: huggingface.co. Saturation forecast: Estimated already saturated. 235 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-78B | 2538.6 | #352 |
| 2 | Qwen 2 VL 72B | 2504.2 | #460 |
| 3 | InternVL3-38B | 2500.7 | #405 |
| 4 | InternVL2.5-78B | 2467.6 | #431 |
| 5 | InternVL3-14B | 2462.3 | #505 |
| 6 | InternVL3-8B | 2422 | #620 |
| 7 | Gemini 2.0 Flash | 2386.1 | #342 |
| 8 | GPT-4.1 (2025-04-14) | 2337.6 | #250 |
| 9 | Qwen 2.5 VL 7B | 2312.1 | #643 |
| 10 | Qwen 2 VL 7B | 2276.3 | #750 |
| 11 | MiniCPM-V-2.6 | 2268.7 | #836 |
| 12 | InternVL2-8B | 2215.1 | #817 |
| 13 | Grok 2 (1212) | 2154.7 | #496 |
| 14 | Moonshot v1 8K | 2151.6 | #617 |
| 15 | Gemini 1.0 Pro | 2148.9 | #933 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mme · How It Works · Data refreshed daily, snapshot 2026-10-20.