OpenVLM MMT-Bench - Color Constancy: leaderboard
The "Color Constancy" subtask of OpenVLM MMT-Bench. OpenCompass OpenVLM evaluation of MMT-Bench: multimodal tasks covering visual recognition, localization, OCR, counting, relation reasoning, visual math, and hallucination detection.
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 27B | 57.1 | #607 |
| 2 | GPT-4.1 Nano | 42.9 | #731 |
| 3 | GPT-4.1 (2025-04-14) | 42.9 | #250 |
| 4 | Qwen 2 VL 7B | 42.9 | #750 |
| 5 | Pixtral-12B | 42.9 | #808 |
| 6 | Qwen 2 VL 2B | 42.9 | #1220 |
| 7 | Gemma 3 12B | 42.9 | #671 |
| 8 | Gemma 3 4B | 42.9 | #1089 |
| 9 | Llama 3.2 11B Instruct | 42.9 | #1126 |
| 10 | Llama 3.2 90B Vision Instruct | 42.9 | #622 |
| 11 | InternVL3-8B | 28.6 | #620 |
| 12 | InternVL3-38B | 28.6 | #405 |
| 13 | InternVL3-78B | 28.6 | #352 |
| 14 | MiniCPM-V-2.6 | 28.6 | #836 |
| 15 | InternVL3-14B | 28.6 | #505 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-color-constancy · How It Works · Data refreshed daily, snapshot 2026-10-20.