Woofun AI reports that Qianwen has launched Qwen-Image-3.0, a third-generation image generation model capable of processing up to 4.5k tokens. The system focuses on generating complex layouts, including newspapers, test papers, and 3x3 information grids, while handling multiple themes, formulas, and bilingual text simultaneously. Text rendering capabilities have been refined to support characters as small as 10px, alongside LaTeX formulas, superscripts, subscripts, and handwritten annotations. The model natively supports 12 languages and over 100 artistic styles, targeting practical applications in presentations, education, and design rather than purely aesthetic outputs.