wavespeed-ai/flux-dev-ultra-fast

Positive prompt words for content generation (required).

Input image used as a reference for generating images (optional).

Mask image: The white area indicates where new pixels need to be generated, while the black area indicates where the original image should be preserved. Used for guided image editing.

The intensity of transformation applied to the reference image, ranging from 0.0 to 1.0. The higher the value, the stronger the transformation.

Resolution of the generated media (widthheight, in pixels). It is recommended that each side’s pixel range be between 256 and 1536. Examples: 10241024, 768*1344.

Number of inference steps, range 1 ~ 50. Higher step count usually results in better quality but takes longer.

The random seed used during generation. A value of -1 indicates random generation. Range: -1 ~ 2147483647.

Guidance Scale, range 0 ~ 20. The higher the value, the stronger the influence of the prompt.

Number of images to generate, range 1 ~ 4.

Output image format, optional values: jpeg, png, webp.

Whether to return the result as a BASE64 string (instead of a URL). Available via API only.

Whether to enable synchronous mode. When set to true, the interface will return only after the image generation and upload are completed, allowing you to directly obtain the image in the response. Available only via API.

Callback interface address (optional). After the task is generated, the system will send a result request to this link. Returning status code 200 indicates successful receipt.