Draw Tasks

← API Reference · 简体中文

Create and query text-to-image tasks with the user-level draw key (see API Keys & Authentication). Tasks run asynchronously: get the uuid, then poll the detail until done.

Create a Draw Task

POST /ext/v1/draw/generation

Parameter	Type	Required	Description
prompt	string	yes	The prompt
negativePrompt	string	no	Negative prompt
size	string	no	Image size (model-dependent, e.g. 1024x1024)
quality	string	no	Quality (where supported)
number	int	no	Number of images, ≥ 1, default 1
model	string	no	Image model name
seed	int	no	Random seed

Response:

{ "uuid": "<task uuid>" }

Get a Draw Task

GET /ext/v1/draw/{uuid}

Only tasks owned by the key's owner are visible; otherwise A_DRAW_NOT_FOUND.

Response Fields (DrawDto)

Field	Description
uuid	Task ID
prompt	The prompt
aiModelName / aiModelPlatform	Model and platform used
interactingMethod	Generation method
isPublic	Public flag
starCount / isStar	Like count and whether the current user liked it
processStatus	Processing status (poll this for completion)
processStatusRemark	Status remark (failure reason etc.)
imageUuids / imageUrls	Generated image IDs and URLs
duration	Elapsed time
dynamicParams	Model dynamic parameters
originalImageUuid/Url, maskImageUuid/Url	Original / guidance images (background generation etc.)
userUuid / userName / createTime	Ownership and time

Example

# Create the task
curl -X POST "http://<host>:9999/ext/v1/draw/generation" \
  -H "Authorization: <draw user-level key>" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "A white owl under moonlight, ink painting style", "number": 2}'

# Poll the detail; take images from imageUrls once processStatus is done
curl "http://<host>:9999/ext/v1/draw/<task uuid>" \
  -H "Authorization: <draw user-level key>"

---

Previous: Knowledge Base Q&A · Next: MCP Services