Alibaba's DAMO Academy has proudly unveiled and open-sourced a cutting-edge text-to-image model named Qwen-Image. Positioned as the pioneering foundational image generation model within the Tongyi Qianwen series, Qwen-Image is built upon the robust MMDiT architecture, boasting an impressive 20 billion parameters. Across numerous evaluations for image generation and editing, Qwen-Image has showcased remarkable performance, particularly distinguishing itself in the rendering of Chinese characters. Upon its debut, Qwen-Image swiftly ascended to the top spot on the prestigious HuggingFace model list.