Ristretto-3B: benchmark results
Provider: Other.
Aggregate Rating 1490 ± 32, rank #2510 of 10916 rated models, from 148 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MMBench V1.1 EN - Image Emotion | 84.4 | Accuracy (%) | 98.2 |
| OpenVLM OCRBench - Scene Text-centric VQA | 95 | Correct Answers (%) | 96.3 |
| OpenVLM MMBench V1.1 EN - Attribute Recognition | 96.6 | Accuracy (%) | 93.6 |
| OpenVLM AI2D - Rock Cycle | 76.1 | Accuracy (%) | 93.3 |
| OpenVLM AI2D - Moon Phase and Equinox | 76.2 | Accuracy (%) | 92.2 |
| OpenVLM MMBench V1.1 CN - Image Style | 82.6 | Accuracy (%) | 91.2 |
| OpenVLM AI2D - Faults and Earthquakes | 78.6 | Accuracy (%) | 90.1 |
| OpenVLM MMBench V1.1 CN - Physical Relation | 68 | Accuracy (%) | 90.1 |
| OpenVLM MMMU - History | 80 | Accuracy (%) | 89.8 |
| OpenVLM MMBench V1.1 EN - Physical Property Reasoning | 74.7 | Accuracy (%) | 89.2 |
| OpenVLM AI2D - Water, Carbon, and Nitrogen Cycle | 75 | Accuracy (%) | 88.7 |
| OpenVLM MMBench V1.1 EN - Physical Relation | 70.7 | Accuracy (%) | 88.7 |
Interactive version: theaggregate.ai/model?slug=ristretto-3b · How It Works · Data refreshed daily, snapshot 2026-11-08.