Create a video

Create an asynchronous task from a text prompt and an optional reference image. The response includes a Video ID used to retrieve status and download the result.

POST/v1/videos

Create asynchronous video generation tasks based on text prompts and optional reference images. The interface accepts Multipart requests compatible with OpenAI Videos and returns a Video ID that can be used to query task status. Different video models may support different durations, sizes, reference images, and extension parameters. Please select the model ID that is visible and supports video generation via the current API Key

Authorization

bearerAuth

AuthorizationBearer <token>

Use Bearer your-api-key in the Authorization Header.

In: header

Request Body

multipart/form-data

Use Multipart to submit prompts, models, and optional reference images. The client should automatically generate the boundary of the Content-Type from the request library.

TypeScript Definitions

Use the request body type in TypeScript.

prompt*string

Describe the video content you want to generate. It is recommended to clarify the subject, actions, scenes, shots, style, and quality requirements.

Length1 <= length

model?string

The video model ID to be used. Please select models that are visible to the current API Key and support video generation; Different models may support different input methods and extension parameters.

Length1 <= length

seconds?string

Video duration, measured in seconds, default value is 4; The actual selectable duration depends on the selected model.

Default"4"

size?string

Output video size, usually in WIDTHxHEIGHT format. The default is 720x1280; available sizes depend on the selected model.

Default"720x1280"

input_reference?file

Reference images uploaded when generating images to videos. Some compatible models do not accept direct file uploads and need to use metadata.img_url to pass image URLs instead.

Formatbinary

metadata?string

Vendor- or model-specific extension parameters must be submitted as JSON strings. Do not treat the fields in these as public parameters supported by all video models. The Wan extension keys include: img_url, first_frame_url, last_frame_url, watermark, prompt_extend, audio, seed, and resolution. The image URL is used for image-to-video production, and the first and last frame URLs are for keyframe video;audio Only applicable to models that support audio generation.

Resolution options include 480P (832x480, 480x832, 624x624), 720P (1280x720, 720x1280, 960x960, 1088x832, 832x1088), and 1080P (1920x1080, 1080x1920, 1440x1440, 1632x1248, 1248x1632). The actual supported range is based on the selected model

Response Body

`application/json`

curl -X POST "https://example.com/v1/videos" \  -F prompt="An astronaut in a spacesuit walking on the moon, high quality, cinematic quality" \  -F model="MODEL_ID" \  -F seconds="4" \  -F size="720x1280"

{
  "id": "video_123",
  "object": "video",
  "model": "MODEL_ID",
  "created_at": 1640995200,
  "status": "processing",
  "progress": 0
}

Authorization

Request Body

Response Body

201application/json

400application/json

401application/json

403application/json

404application/json

500application/json

`application/json`

`application/json`

`application/json`

`application/json`

`application/json`

`application/json`