Qwen Image 2.1 – Image to Image is Alibaba’s compact image generation and editing model designed to transform existing images using natural-language instructions and reference images. It combines image understanding with generation, allowing users to make broad visual transformations, preserve important details, or perform targeted edits. Kie.ai supports up to 10 reference images for this workflow. The model supports 1K and 2K output, multiple aspect ratios, transparent backgrounds, prompt enhancement, and optional masking for localized edits. Its compact 7B visual-generation architecture is designed for efficient inference while supporting high-quality image generation and editing.