Qwen-Image
Qwen-Image is a 20B MMDiT image foundation model, released with code and weights, that claims leading results on complex text rendering, especially Chinese.
Progressive curriculum from simple to complex text rendering plus a dual-encoding scheme (semantic and reconstructive) for editing consistency. Strong in logographic scripts such as Chinese while keeping English quality.
- Date
- Monday, 4 August 2025
- Lab
- Alibaba (Qwen)
- Kind
- open-weights
- Access
- open weights
Figures
| Measure | Value | Measured by |
|---|---|---|
| Parameters | 20B (MMDiT) | authors |
Date is arXiv v1. Weights licence (believed Apache 2.0) not verified from the paper page; check the GitHub repo.
Sources
This record was checked against its sources on 6 October 2026. How we check