Qwen 3 VL 30B A3B: benchmark results
Provider: Alibaba. Released 2025-10-04. Access: Open.
Aggregate Rating 1631 ± 4, rank #351 of 1656 rated models, from 741 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EarthSpatialBench (CoT) - Angle Estimation | 54.99 | Mean absolute error in degrees of the estimated azimuth betw | 100 |
| EarthSpatialBench (Direct) - Direction Binary Classification | 71 | Macro-averaged F1 (%) on EarthSpatialBench yes/no direction | 100 |
| HAERAE-Vision - Mathematics | 68.69 | Checklist Score (%; original under-specified queries, GPT-5- | 100 |
| IKEA-Bench (Text Only) - Step Verification | 62 | Accuracy (%) of binary step verification (judge whether a vi | 100 |
| IKEA-Bench (Visual+Text) - Step Verification | 65.7 | Accuracy (%) of binary step verification (judge whether a vi | 100 |
| MPCI-Bench - Trace Tier Probing | 90.4 | Probe Accuracy (%) | 100 |
| MinCU | 51.76 | Grounded F1 (%; overall on 2,000 image pairs, 500 per change | 100 |
| MinCU - Relation | 49.37 | Grounded F1 (%; relation changes (an entity relocated with i | 100 |
| OVIBench (Multiple Choice) - Cancellation | 65.03 | Answer-selection accuracy (%; choosing, among four candidate | 100 |
| PlantXpert - Quantitative Challenges | 55.26 | Accuracy (%) on the quantitative reasoning questions on the | 100 |
| PlantXpert - Remote Sensing and Imaging | 90.91 | Accuracy (%) on the remote sensing and imaging domain questi | 100 |
| PlantXpert - Soybean | 71.28 | Accuracy (%) on the 296 soybean questions on the expert-anno | 100 |
Interactive version: theaggregate.ai/model?slug=qwen-3-vl-30b-a3b · How It Works · Data refreshed daily, snapshot 2026-10-20.