granite-vision-3.1-2B-preview: benchmark results
Provider: IBM. Released 2025-01-31. Access: Open.
Aggregate Rating 1390 ± 31, rank #6534 of 10916 rated models, from 148 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM AI2D - Atom Structure | 87.5 | Accuracy (%) | 85.2 |
| OpenVLM MMBench V1.1 EN - Attribute Recognition | 94.4 | Accuracy (%) | 78.8 |
| OpenVLM MMBench V1.1 EN - Image Emotion | 80 | Accuracy (%) | 76 |
| OpenVLM AI2D - Volcano | 93.8 | Accuracy (%) | 75.8 |
| OpenVLM MathVista - MWP | 73.7 | Accuracy (%) | 71.1 |
| OpenVLM MMBench V1.1 EN - Identity Reasoning | 97.4 | Accuracy (%) | 71 |
| OpenVLM MathVista - FQA | 66.9 | Accuracy (%) | 70.2 |
| OpenVLM MathVista - STA | 74.1 | Accuracy (%) | 68 |
| OpenVLM MathVista - ARI | 59.8 | Accuracy (%) | 67.3 |
| OpenVLM OCRBench - Handwritten Mathematical Expression Recognition | 59 | Correct Answers (%) | 65 |
| OpenVLM MMBench V1.1 EN - Physical Relation | 61.3 | Accuracy (%) | 64.8 |
| OpenVLM MathVista - VQA | 49.7 | Accuracy (%) | 61.9 |
Interactive version: theaggregate.ai/model?slug=granite-vision-3-1-2b-preview · How It Works · Data refreshed daily, snapshot 2026-11-08.