Qwen Image 3.0 renders an image from a description or edits from one to three of your own references, at the 1K or 2K tier, with PNG or JPEG output.
Use an uploaded reference to guide the subject, product or layout, and describe the changes you want. Review the result before using it.
Plan your prompt controls. Prompt rewriting is switched on by default, so a short description can be enriched with visual detail, and you can switch it off when the exact wording matters. An optional negative prompt and an optional seed cover the rest of the repeat work.
Choose a tier (1K or 2K), a shape, an output format and, optionally, a negative prompt and a seed. Prompt rewriting is on by default. Every reference image is quoted separately, starting with the first, so the quote rises with the number you upload.
Upload between one and three images, for example a product, a colour reference and a setting, and describe how they should combine in one render.
Eight shapes from square to 21:9, and the same list stays selectable when you start from a reference instead of a description.
Leave it on to let the model add visual detail to a short prompt, or turn it off so your wording reaches the model unchanged.
Draft from a description, then move to reference editing once a version is close and the same subject is needed in new settings.
Render the same idea in square, portrait and 21:9 so one concept covers feed, story and header placements.
Leave prompt rewriting on to expand a few words into a fuller scene, then tighten the wording yourself once the direction is right.
Use a button on this page. It opens the right workspace with this model already selected, in the language you are reading now.
Write what should happen in the shot or image. Concrete subjects, framing and action give the model more to work with than short labels.
Choose length or quality tier, shape and sound where offered, then read the credit quote for those exact settings before you generate.
Generation uses account credits. Cost depends on the model, quality and output size or duration. Reference images, video or audio may add charges in some modes. Check the workspace quote for your selected inputs and settings before generating; the pricing page lists credit packages.
See credit packagesBetween one and three per editing task. Each reference is quoted on top of the base amount, from the first one, so the quote rises with the number you upload.
It can. The control is on by default and lets the model enrich a short description with visual detail. Switch it off when exact wording, such as text that must appear on the image, has to reach the model unchanged.
For this model both tiers are quoted at the same amount, so the choice is about the quality setting you want rather than the price.
Uploads can be JPEG, PNG or WebP, up to 10 MB per image. Finished files are PNG or JPEG, selected before you generate.
The seed field is optional and takes an integer. We do not promise that repeating a seed reproduces an identical image, so compare the results instead of assuming a match.
Other models on this platform that do a similar job, or that pair well with this one.
Generate images from descriptions or edit with one to three references. Qwen Image 3.0 Pro offers 1K and 2K tiers, eight aspect ratios, prompt rewriting and PNG or JPEG output.
Read the guideThe FLUX 3 family product available on this platform: text generation and reference editing at 1K, 2K or 4K, with PNG, JPEG and WebP output.
Read the guideNano Banana 2.1 covers both workflows on one model: images from a description, or edits driven by up to ten of your own reference images, at 1K, 2K or 4K.
Read the guide