MMEB Image: leaderboard
MMEB image split evaluating multimodal embedding models on image classification, QA, retrieval, and visual grounding tasks.
Metric: Image Overall Score (%). Source: huggingface.co. 101 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Ovis-VL-Embedding-9B | 81.69 | |
| 2 | WeMM-Embedding-9B | 80.99 | |
| 3 | QQMM-embed-v4 | 79.81 | |
| 4 | WeMM-Embedding-4B | 79.78 | |
| 5 | Wianmu-5B-v0 | 79.73 | |
| 6 | octen-vl-embedding-large | 79.65 | |
| 7 | Wanderboat-Embedding-2B | 78.97 | |
| 8 | DME-Large | 78.91 | |
| 9 | Wianmu-3B-v0 | 78.67 | |
| 10 | WeMM-Embedding-2B | 78.63 |
Interactive version: theaggregate.ai/benchmark?slug=mmeb-image · How It Works · Data refreshed daily, snapshot 2026-10-20.