Vintern-1B-v2: benchmark results
Provider: VinAI. Released 2024-08-14. Access: Open.
Aggregate Rating 1263 ± 38, rank #9144 of 10916 rated models, from 145 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MMMU - Diagnostics and Laboratory Medicine | 43.3 | Accuracy (%) | 75.3 |
| OpenVLM MMMU - Materials | 40 | Accuracy (%) | 71.7 |
| OpenVLM MMMU - Mechanical Engineering | 36.7 | Accuracy (%) | 54.8 |
| OpenVLM HallusionBench - aAcc | 61.7 | Accuracy (%) | 49.6 |
| OpenVLM OCRBench - Key Information Extraction | 73 | Correct Answers (%) | 44.4 |
| OpenVLM OCRBench - Handwritten Mathematical Expression Recognition | 15 | Correct Answers (%) | 43.1 |
| OpenVLM AI2D - Water, Carbon, and Nitrogen Cycle | 56.8 | Accuracy (%) | 39.4 |
| OpenVLM MMMU - Physics | 30 | Accuracy (%) | 34.5 |
| OpenVLM OCRBench | 62.7 | Score (normalized) | 33.8 |
| OpenVLM MathVista - GPS | 42.3 | Accuracy (%) | 32.5 |
| OpenVLM HallusionBench - qAcc | 28.1 | Accuracy (%) | 32.2 |
| OpenVLM OCRBench - Scene Text-centric VQA | 81 | Correct Answers (%) | 31.9 |
Interactive version: theaggregate.ai/model?slug=vintern-1b-v2 · How It Works · Data refreshed daily, snapshot 2026-11-08.