Alibaba unveils Qwen-Image-2.1: a 7b-parameter model for image generation and editing

Alibaba’s Qwen team has launched Qwen-Image-2.1, a unified text-to-image generation and editing model with 7B parameters. The model integrates both generation and editing capabilities into a single framework, reducing the size to about a third of its predecessor. It supports features like multi-reference editing, local edits, and transparent RGBA output. Deployment is available for research and evaluation, with commercial use requiring a separate license. The model’s performance on Qwen’s internal benchmark, Qwen-Image-Bench, scores 60.28, surpassing other open-weight models like Nano Banana 2.0 and FLUX 2 Max. The release also includes two prompt-rewriting models fine-tuned for text-to-image tasks. The GitHub repository provides detailed technical information and installation instructions for various frameworks and platforms.