H2OVL-2B: benchmark results
Provider: H2O.ai. Released 2024-09-27. Access: Open.
Aggregate Rating 1368 ± 37, rank #7217 of 10916 rated models, from 439 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MMT-Bench - 3D Indoor Recognition | 60 | Score (%) | 99.5 |
| OpenVLM MMT-Bench - Image Matting | 45 | Score (%) | 99.3 |
| OpenVLM MMT-Bench - Rotated Object Detection | 77.8 | Score (%) | 96.4 |
| OpenVLM SEEDBench IMG - Text Understanding | 77.4 | Accuracy (%) | 95.1 |
| OpenVLM OCR-VQA | 70.3 | Accuracy (%) | 94 |
| OpenVLM MME - Scene | 166.2 | Score | 92.9 |
| OpenVLM MMT-Bench - Body Emotion Recognition | 55 | Score (%) | 89.6 |
| OpenVLM MMT-Bench - Color Contrast | 70 | Score (%) | 87.1 |
| OpenVLM Q-Bench - Yes-or-No (In-context Others) | 90 | Accuracy (%) | 87 |
| OpenVLM MMMU - Mechanical Engineering | 50 | Accuracy (%) | 83.7 |
| OpenVLM MMT-Bench - MIA | 65 | Score (%) | 83.5 |
| OpenVLM MMT-Bench - Deepfake Detection | 60 | Score (%) | 83 |
Interactive version: theaggregate.ai/model?slug=h2ovl-2b · How It Works · Data refreshed daily, snapshot 2026-11-08.