Asynchronous generation

Create an image-generation task in the background, save its task ID, and query it for status and final output.

Create an asynchronous task when image generation should continue independently of the original connection. Save the returned id, poll its status with increasing intervals, and download the signed result URL promptly after success.

Authorization and request format

Authorizationstring · headerrequired

Pass Authorization: Bearer $WYLON_API_KEY.

Content-Typestring · headerrequired

Send the request body as application/json.

Create a task

Create parameters

typestringrequired

Use image. If omitted, the API treats the request as a video task.

modelstringrequired

A complete wylon model ID, such as Qwen/Qwen-Image-2512.

promptstringrequired

Describe the subject, scene, composition, lighting, style, and text to generate.

negative_promptstringoptional

Describe content to exclude. Support depends on the model.

seedintegeroptional

Controls generation randomness when supported by the model.

sizestringoptional

Output dimensions in widthxheight format.

ninteger1

Number of images to generate.

output_formatstringpng

Requested output file format.

output_compressionintegeroptional

Requested compression level. It has no effect on PNG output by default.

backgroundstringoptional

OpenAI-compatible background option. It has no effect in V1 by default.

qualitystringoptional

OpenAI-compatible quality option. It has no effect in V1 by default.

stylestringoptional

Requested style preset. Supported values depend on the model.

userstringoptional

A caller-supplied end-user identifier. It is not used for authentication.

user_idstringoptional

Compatibility field accepted by the API but not currently stored or used.

moderationstringoptional

Requested moderation mode. It cannot disable platform safety controls.

streambooleanfalse

V1 does not support partial-image streaming; omit this field or pass false.

partial_imagesinteger0

V1 does not support partial images; omit this field or pass 0.

Current model limits

For Qwen/Qwen-Image-2512 only:

Parameter	Supported values	Default
`size`	`1328x1328`, `1664x928`, `928x1664`, `1104x1472`, `1584x1056`	`1664x928`
`n`	`1`	`1`

V1 adds an “AI-generated” watermark and does not support partial-image streaming.

Query a task

Send a GET request with the returned task ID. Add detail=true to include the effective image_parameters. Poll at a reasonable interval and stop at a terminal status.

Status	Meaning	Action
`queued`	Accepted and waiting	Query again later
`running`	Generating	Query again later
`succeeded`	Completed	Download the image
`failed`	Generation failed	Read `error`
`cancelled`	Cancelled while queued	Stop polling
`expired`	Task or result expired	Create a new task if needed

Read and download the result

When status=succeeded and file_status=available, the signed URL is in data[0].url and its expiry time is in data[0].url_expires_at. Download it promptly and persist the file in your own storage if needed. After expiry, file_status changes to expired, and the URL is not reissued.

Error handling

For network errors or 5xx responses after task creation, query the original task before submitting another request. This prevents duplicate generation and charges.