Wan 3.0 Video

Model ID alibaba/wan-3.0-video

Overview

One endpoint selects text-to-video, keyframe image-to-video, or reference-to-video from the media fields you send. Keyframe and reference inputs cannot be mixed.

Alibaba · text-image-reference-to-video

Input Modes

The media fields actually sent in the request select which capability runs:

Important: Keyframe mode (image/end_image) and reference mode (images/videos/audios) are mutually exclusive. Do not mix fields from both groups in one request.

Endpoint

POST /v1/video/generations

Authentication

Authorization: Bearer <AIAPIAIAPI_API_KEY>

Request Body

model

Type: string · Required: Yes · Nullable: No

Model ID used to route the request.

prompt

Type: string · Required: No · Nullable: No

Text prompt. Use it alone for text-to-video; in reference mode, refer to media with independent 1-based labels such as Image 1, Video 1, and Audio 1.

image

Type: string · Required: No · Nullable: No

First-frame image for keyframe mode. Cannot be combined with images, videos, or audios.

end_image

Type: string · Required: No · Nullable: No

Last-frame image for keyframe mode. May be sent with or without image, but cannot be combined with reference-mode fields.

images

Type: array · Required: No · Nullable: No

Reference-mode images. Up to 10; reference them in prompt as Image 1, Image 2, and so on.

videos

Type: array · Required: No · Nullable: No

Reference-mode videos. Up to 5 HTTPS URLs; reference them in prompt as Video 1, Video 2, and so on.

audios

Type: array · Required: No · Nullable: No

Reference-mode audio clips. Up to 5 HTTPS URLs; reference them in prompt as Audio 1, Audio 2, and so on.

audio_enable

Type: boolean · Required: No · Nullable: No

Whether to generate audio together with the video. Defaults to true.

duration

Type: integer · Required: No · Nullable: No

Video duration in seconds from 2 to 30, or -1 for smart duration.

resolution

Type: string · Required: No · Nullable: No

Output resolution. Accepts 480p, 720p, or 1080p case-insensitively; defaults to 1080p.

aspect_ratio

Type: string · Required: No · Nullable: No

Output aspect ratio. Defaults to adaptive; also supports 16:9, 4:3, 1:1, 3:4, and 9:16.

seed

Type: integer · Required: No · Nullable: No

Random seed from 0 to 2147483647 for reproducible output. Omit it for a random seed.

Request Example

curl --request POST \
  --url https://aiapiaiapi.com/v1/video/generations \
  --header 'Authorization: Bearer $AIAPIAIAPI_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
      "model": "alibaba/wan-3.0-video",
      "prompt": "A cinematic camera movement through a forest",
      "duration": 5,
      "resolution": "1080p"
    }'

JSON

{
  "model": "alibaba/wan-3.0-video",
  "prompt": "A cinematic camera movement through a forest",
  "duration": 5,
  "resolution": "1080p"
}

Response

{
  "id": "vid_01K2ABC123",
  "object": "video.generation",
  "model": "alibaba/wan-3.0-video",
  "status": "queued",
  "progress": null,
  "outputs": [],
  "created_at": 1786960800,
  "started_at": null,
  "completed_at": null,
  "error": null
}

Task Status

Use the id from the create response as task_id:

GET /v1/video/generations/{task_id}

{
  "id": "vid_01K2ABC123",
  "object": "video.generation",
  "model": "alibaba/wan-3.0-video",
  "status": "completed",
  "progress": 100,
  "outputs": [
    { "type": "video", "url": "https://cdn.example.com/videos/vid_01K2ABC123.mp4" }
  ],
  "created_at": 1786960800,
  "started_at": 1786960805,
  "completed_at": 1786960900,
  "error": null
}

Error Responses

Validation errors return HTTP 400 with a stable error code.

{
  "error": {
    "type": "invalid_request_error",
    "code": "unsupported_value",
    "message": "The parameter value is not supported by this model.",
    "param": "duration"
  }
}