OpenVLM MMStar - Coarse Perception: leaderboard
OpenCompass OpenVLM MMStar subcategory score for coarse visual perception questions.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around August 2028. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2 VL 72B | 74.4 | #460 |
| 2 | GPT-5 | 74 | #98 |
| 3 | GPT-5 Nano | 74 | #425 |
| 4 | Gemini 1.5 Pro (002) | 73.6 | #349 |
| 5 | GPT-4.1 (2025-04-14) | 73.6 | #250 |
| 6 | Gemini 1.5 Pro | 73.6 | #445 |
| 7 | GPT-4o ChatGPT | 73.6 | |
| 8 | InternVL3-8B | 73.2 | #620 |
| 9 | Gemini 2.5 Pro | 72.4 | #152 |
| 10 | GPT-4.1 Mini | 72.4 | #355 |
| 11 | Gemini 1.5 Flash | 72.4 | #618 |
| 12 | Gemini 2.0 Flash | 72 | #342 |
| 13 | Qwen 2 VL 7B | 72 | #750 |
| 14 | GPT-5 Mini (2025-08-07) | 72 | #172 |
| 15 | Qwen 2.5 VL 7B | 72 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmstar-coarse-perception · How It Works · Data refreshed daily, snapshot 2026-10-20.