Generate an image with FLUX 3

Non-blank text describing the image to generate or the edit to apply.

To place elements, name each one as <id> in the text, then end the prompt with a JSON list of rows, one per element. Each bbox is [top, left, bottom, right] on a 0 to 1000 grid, so [0, 0, 500, 500] is the top-left quarter of the output:

A black silhouette of a running person <runner_1>
on a flat chartreuse background <background_1>.
[
  {"id": "background_1", "bbox": [0, 0, 1000, 1000], "desc": "chartreuse field"},
  {"id": "runner_1", "bbox": [150, 150, 850, 850], "desc": "running silhouette"}
]

In an edit, rows use from, src_bbox, and tgt_bbox instead of bbox. There is no separate box field, and an entities field returns 422. For full requests, choose Layout with boxes or Edit with boxes in the request examples. Layout and editing lists every row field.

One reference image or a list of 1 to 10 images, each as an HTTP(S) URL or a base64 string. A data URI prefix is accepted. Each image must be at least 256 × 256 pixels and at most 16 megapixels; each base64 payload must be at most 20 MB. Inputs over 16 megapixels are rejected with 400, inputs below 256 × 256 pixels with 422.

With auto, output follows the first reference image's aspect ratio, or uses 1:1 when there is no reference image.

Choose the output size. Check output_mp in the submit response for the exact size in megapixels. Generating at 4k can take several minutes.

Input and output moderation tolerance; 0 is strictest.

When true, the model can research the prompt with web search and image search before it generates. Set false to turn both off.

Task identifier.

Regional polling URL. Poll this URL with the x-key header rather than constructing a URL.

Cost in credits for this request

Input megapixels (2 decimal places)

Output megapixels (2 decimal places)