GPT-IMG
Instructions, not re-descriptions
Write what should change — "make the jacket red", "remove the sign", "put this on a marble counter" — instead of re-describing the whole scene and hoping the model reconstructs it.
GPT-IMGQwen Image Edit: change a photo by describing the change
Alibaba built this model for instruction-based editing rather than generation from scratch. Upload a picture, say what should be different, and keep the rest intact.
Upload an image and describe the change. Alibaba's Qwen Image Edit rewrites objects, backgrounds, and styling in place for 15 credits, no install required.
Flat 15 credits per editOne reference image, described in plain languageOpen-weights Qwen model, hosted for you
0 / 20,000
Image
Upload at least one reference image before generating.
An editor, not a generator
Qwen Image Edit only runs with an image attached. That constraint is what makes it good at leaving the rest of the frame alone.
GPT-IMG
The untouched parts stay untouched
Because it starts from your pixels rather than from noise, faces, poses, and layout survive edits that would drift on a text-to-image model.
GPT-IMG
One image at a time
It takes exactly one input. For multi-reference composition — a subject from one photo into a scene from another — use FLUX 2 Pro or Nano Banana Pro.
GPT-IMG
Five output shapes
Square, 4:3, 3:4, 16:9, and 9:16. There is no 2:3 or 3:2 here, and no 2K or 4K tier.
Keep exploring GPT-IMG
One account, the full image generation workflow.
Choose Your Plan
Upgrade for larger monthly credits, faster generation, and commercial-ready output.
FAQs
Practical details for this generator page.
Start with Qwen Image Edit: change a photo by describing the change
Use the prompt panel above with the same account, credits, and history workflow.