Qwen3-VL Instruct (4B): benchmark results
Provider: Alibaba. Released 2025-10-15. Access: Open.
Aggregate Rating 1584 ± 49, rank #1191 of 10916 rated models, from 28 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Robo-Reward-Bench - Utokyo Xarm Bimanual Converted Externally To Rlds | 1 | Absolute Error | 100 |
| HELM Robo-Reward-Bench - Berkeley Autolab Ur5 | 0.68 | Absolute Error | 75 |
| HELM Robo-Reward-Bench - Robo Arena | 0.93 | Absolute Error | 70.8 |
| HELM Robo-Reward-Bench - Ucsd Kitchen Dataset Converted Externally To Rlds | 1.06 | Absolute Error | 68.8 |
| HELM Robo-Reward-Bench - Taco Play | 1.16 | Absolute Error | 54.2 |
| HELM Robo-Reward-Bench - Iamlab Cmu Pickup Insert Converted Externally To Rlds | 0.81 | Absolute Error | 52.1 |
| HELM Robo-Reward-Bench - Berkeley Rpt Converted Externally To Rlds | 0.93 | Absolute Error | 50 |
| HELM Robo-Reward-Bench - Cmu Play Fusion | 1.01 | Absolute Error | 45.8 |
| HELM Robo-Reward-Bench - Kaist Nonprehensile Converted Externally To Rlds | 1.09 | Absolute Error | 41.7 |
| HELM Robo-Reward-Bench - Viola | 0.6 | Absolute Error | 37.5 |
| HELM Robo-Reward-Bench - Dlr Edan Shared Control Converted Externally To Rlds | 0.89 | Absolute Error | 35.4 |
| HELM Robo-Reward-Bench - Bridge | 1.14 | Absolute Error | 33.3 |
Interactive version: theaggregate.ai/model?slug=qwen3-vl-instruct-4b · How It Works · Data refreshed daily, snapshot 2026-11-08.