Qwen-Image (20B MMDiT)
20B MMDiT image model with native text rendering for English and Chinese, opened under Apache 2.0 with a technical report.
Open image foundation model focused on multi-line and paragraph-level text rendering plus consistent editing through a multi-task training scheme. Followed by Edit on 2025-08-19 and, through 2026-09, by Edit-2509, 2511, Layered, 2512, 2.0, 3.0 and 2.1.
- Date
- Monday, 4 August 2025
- Lab
- Alibaba (Qwen)
- Kind
- open-weights
- Access
- open weights
Figures
| Measure | Value | Measured by |
|---|---|---|
| Technical report | arXiv 2508.02324 Submitted 2025-08-04; model about 20B parameters | company |
Blog dated 2025-08-04 (Hugging Face repo 2025-08-02). Benchmark claims ('outperforms existing models') are Alibaba's and no table was reproduced.
Sources
This record was checked against its sources on 6 October 2026. How we check