From the model card
Fields below are copied from the tags and counters on the HuggingFace repository Qwen/Qwen-Image-Edit-2509 at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- Qwen
- Pipeline tag
- image-to-image
- Library
- Diffusers
- Weight formats
- safetensors
- License tag
apache-2.0— read the license file in the repo before relying on it- Language tags
- English (en), Chinese (zh)
- Papers cited
- arXiv:2508.02324
- Downloads (HF counter at last fetch)
- 443,650
- Likes (HF counter at last fetch)
- 1,238
- Model card
- https://huggingface.co/Qwen/Qwen-Image-Edit-2509
Use cases
- Text-guided object replacement in existing photos
- Style transfer via natural language editing instructions
- Bilingual (EN/ZH) image editing pipelines
- Automated product photo retouching from text prompts
- Research into instruction-following image manipulation
Pros
- Bilingual instruction support (English and Chinese) uncommon in image edit models
- Apache 2.0 license permits commercial use without restrictions
- Diffusers-compatible, enabling integration with existing HuggingFace pipelines
- Over 520k downloads indicates active community validation
- Backed by a peer-reviewed arxiv paper (2508.02324) describing the methodology
Cons
- Instruction-following quality degrades on fine-grained spatial edits (e.g., move object left 10px)
- No dedicated fine-tuning dataset disclosed, limiting domain adaptation guidance
- Diffusers dependency adds overhead compared to single-file inference runtimes
- English/Chinese only — other languages not supported
- Image quality for complex multi-object edits may require prompt engineering iterations
Tags
diffuserssafetensorsimage-to-imageenzharxiv:2508.02324license:apache-2.0diffusers:QwenImageEditPlusPipelineregion:us