Built on Qwen2.5-VL-72B — one of the most capable open vision-language models — delivering deep image understanding and nuanced editing control that smaller models simply can't match.

Edit text within images in both English and Chinese with high accuracy. Perfect for localization workflows, bilingual content, and markets that require both language scripts in a single design.

Apply multiple edits sequentially to the same image without quality degradation. Build complex transformation pipelines — adjust colors, then add elements, then change styles — all in one coherent workflow.

Two editing modes in one model: semantic editing changes concepts and content (swap objects, change subjects), while appearance editing transforms style, color, and texture while preserving structure.

Qwen Image Edit Plus achieves state-of-the-art results on GEdit-Bench and ImgEdit — the leading image editing benchmarks — demonstrating superior instruction following and edit quality versus alternatives.

Built on an Apache 2.0 open model — providing transparency, customizability, and freedom from proprietary lock-in. Enterprise teams can audit, fine-tune, and self-host with full flexibility.