Qwen open-sources Qwen-Image-2.1: a 7B model unifying generation and editing with native transparency

Qwen 开源 Qwen-Image-2.1：7B 统一生成与编辑并原生支持透明图像

Qwen released Qwen-Image-2.1, a 7B model that merges text-to-image generation and image editing into one lightweight system. It natively handles transparent images—generating them from prompts, editing layers, and extracting subjects from photos as RGBA assets. Editing supports up to 10 reference images, local edits, and identity preservation. A mixed-granularity attention design with KV cache reuse cuts inference cost for multi-image tasks. The model is open-sourced on GitHub, Hugging Face, and ModelScope.

Why it matters: Qwen open-sources a 7B unified image model with native transparency — a real differentiator, not a benchmark flex. Editing supports up to 10 reference images, which is practically useful. Score held back because the post doesn't disclose inference latency or VRAM requirements,...