This model is deprecated. Use Qwen Image 3.0 instead.
Image
Qwen Image 2512
Use it ↗Deprecated, replaced by Qwen Image 3.0. Generates and transforms images up to 2048px with photoreal humans, native text, up to 6 LoRAs and a Go Fast option.
Released in December 2025, Qwen Image 2512 is Alibaba's most advanced open-source text-to-image model, competing with alternatives like Google's Nano Banana Pro. It prioritizes final image quality and realism over raw speed, reducing "AI-generated" artifacts for photorealistic results, especially in human portraits. This makes it highly effective for creating character art, product visualizations, and marketing materials with integrated, readable text in both Chinese and English. This precision is achieved through an advanced 20B parameter architecture that renders details like skin texture and hair with exceptional accuracy.