Qwen 2 VL 7B Instruct: benchmark results
Alibaba's Apache-2.0 7B vision-language model (August 2024) handling image, multi-image, and video input. Provider: Alibaba. Released 2024-08-29. Access: Open.
Aggregate Rating 1485 ± 7, rank #826 of 1656 rated models, from 303 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CogRail RailThreat (Cog-MRSI, Type-I Visual Prompt) | 58.61 | Overall F1 (%; harmonic mean of macro precision and macro re | 100 |
| OODBench (OOD-Simple, CoT) | 71.61 | Accuracy (%) on balanced yes/no existence questions ('Does t | 100 |
| V-DyKnow - Visual Entity Recognition | 91 | Accuracy (%) of naming the entity depicted in each V-DyKnow | 100 |
| FewMMBench (4-Shot, Random) | 80.2 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 92 |
| FewMMBench (4-Shot, Similar + CoT) | 77.2 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 92 |
| FewMMBench (4-Shot, Similar) | 80.3 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 92 |
| FewMMBench (8-Shot, Random) | 80.4 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 92 |
| FewMMBench (8-Shot, Similar + CoT) | 77.1 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 92 |
| FewMMBench (8-Shot, Similar) | 80 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 92 |
| OODBench (OOD-Hard) | 66.77 | Accuracy (%) on balanced yes/no existence questions ('Does t | 88.9 |
| OODBench (OOD-Simple) | 73.01 | Accuracy (%) on balanced yes/no existence questions ('Does t | 88.9 |
| TempCompass | 67.85 | Avg All (%) | 85.7 |
Interactive version: theaggregate.ai/model?slug=qwen-2-vl-7b-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.