Image, video, and audio tasks

Generation models use different parameters, although most share a task endpoint. Copy the example from the model page and follow its required fields, types, and options.

1. Create a task

For an accepted asynchronous task, save data.taskId. Some operations return results immediately; follow the actual response.

{"code":200,"msg":"success","data":{"taskId":"task_example"}}

2. Query the task

The task is finished when data.state is success or fail. For some models, data.resultJson is a JSON string containing result URLs; others may use data.response or another structure. Check the actual response.

Input media and previous tasks

Use complete media URLs accessible to the server and follow the model's supported types and size limits. Continuation, conversion, or resource-reference operations can require an earlier task ID or resource ID. Those IDs must belong to tasks accessible to your account; some endpoints also require originTaskIds.

Tasks may consume quota. Check the model, parameters, and price before submission. Example URLs and task IDs are placeholders.

Image, video, and audio tasks

1. Create a task#

2. Query the task#

Input media and previous tasks#

1. Create a task

2. Query the task

Input media and previous tasks