Qwen-Image
Alibaba's open image model built for text rendering, strongest in Chinese
Visit website →About
Qwen-Image is the image generation and editing model from Alibaba's Qwen team. The original release in August 2025 was published under the Apache 2.0 licence with a stated focus on text rendering, especially Chinese, and editing variants followed. A newer generation launched in February 2026 through Alibaba's cloud, without public weights at launch.
Its technical report is one of the clearest explanations of why Arabic goes wrong: it says most systems are trained predominantly on English or Chinese glyphs and therefore struggle with accurate characters, consistent spacing and the correct reading order in other scripts. The report claims substantially improved multilingual rendering but lists no languages, and its published benchmarks are Chinese and English only. Arabic is not named and not tested.
Pricing in October 2026: the open-weight release is free to download and run; cloud access is billed through Alibaba's platform.
AI-assisted draft, human-reviewed before publishing — see how we choose & review tools.
Why we picked it
Qwen-Image matters to Arabic users for its diagnosis more than its output: its own report names the training-data imbalance that breaks Arabic across the whole industry. It is also the only open-weight model built around text rendering, so it is where a Gulf team would start a fine-tune.
Featured in our guides
Related tools
- 1 ArabicDesign.ai Arabic-first calligraphy generator with PNG and SVG export Freemium
- 2 Magnific Freepik's AI suite under its new name: many models, one credit balance Paid
- 3 Seedream ByteDance image model with the most explicit Arabic text claim we found Custom
- 4 Stable Diffusion The open-weight image model family that runs on your own hardware Freemium
- 5 Leonardo.Ai Creator-oriented image platform with a daily free token allowance Freemium
- 6 Recraft Design-focused generator with vector output you can edit afterwards Freemium