Qwen2-VL Instruct (72B): benchmark results
Provider: Alibaba. Access: Open.
Aggregate Rating 1646 ± 22, rank #644 of 10916 rated models, from 136 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM VHELM - Blink Reasoning - IQ Test | 52.67 | Prefix Quasi-Exact Match (%) | 100 |
| HELM VHELM - Hateful Memes | 77.46 | PEM | 100 |
| HELM VHELM - Mme | 93.52 | PEM | 100 |
| HELM VHELM - Mme - Artwork | 88.75 | Prefix Quasi-Exact Match (%) | 100 |
| HELM VHELM - Seed Bench - Visual Reasoning | 84.29 | Prefix Quasi-Exact Match (%) | 100 |
| HELM VHELM - Toxicity | 100 | Mean win rate | 100 |
| HELM VHELM - Unicorn | 91.2 | PEM | 100 |
| HELM VHELM - Unicorn - OODCV-VQA | 84.8 | Prefix Quasi-Exact Match (%) | 100 |
| HELM VHELM - Unicorn - Sketchy-VQA | 97.6 | Prefix Quasi-Exact Match (%) | 100 |
| HELM VHELM - Vqa Base | 89.9 | PEM | 100 |
| HELM VHELM - Vqa Dialect | 87.7 | PEM (Fairness) | 100 |
| HELM VHELM - Mmmu - Materials | 63.33 | Prefix Quasi-Exact Match (%) | 98.9 |
Interactive version: theaggregate.ai/model?slug=qwen2-vl-instruct-72b · How It Works · Data refreshed daily, snapshot 2026-11-08.