OpenVLM MME - Scene: leaderboard
OpenCompass OpenVLM MME subcategory score for recognizing and understanding scenes.
Metric: Score. Source: huggingface.co. Saturation forecast: Estimated already saturated. 235 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Pixtral-12B | 169.5 | #808 |
| 2 | InternVL3-38B | 165.5 | #405 |
| 3 | Qwen 2 VL 72B | 164 | #460 |
| 4 | Qwen 2 VL 7B | 162.5 | #750 |
| 5 | InternVL3-8B | 160.8 | #620 |
| 6 | Qwen 2 VL 2B | 159 | #1220 |
| 7 | InternVL3-78B | 158.8 | #352 |
| 8 | Qwen 2.5 VL 7B | 158.5 | #643 |
| 9 | Gemini 1.5 Flash | 158 | #618 |
| 10 | Moonshot v1 8K | 158 | #617 |
| 11 | GPT-4.1 Nano | 157.8 | #731 |
| 12 | Grok 2 (1212) | 157.2 | #496 |
| 13 | InternVL3-14B | 157 | #505 |
| 14 | Aquila-VL-2B | 157 | #1238 |
| 15 | InternVL2.5-78B | 156 | #431 |
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-scene · How It Works · Data refreshed daily, snapshot 2026-10-20.