Wan 3.0 Video Generation

Submit an asynchronous Alibaba Cloud Wan 3.0 video generation task. Supports text-to-video, image-to-video (first frame / first-last frame), and reference-to-video.

This capability uses the top-level media array as multimodal input. Different combinations of media.type enable different capabilities:

• Text-to-video: Only pass prompt, no media.
• First frame to video: Pass first_frame type media.
• First-last frame to video: Pass both first_frame and last_frame type media.
• Reference to video: Pass reference_image, reference_video, or reference_audio type media.

Note: reference_xx / file / link types and first_frame / last_frame types are mutually exclusive and cannot be mixed in the same request.

After submission, query the task status via GET /v1/videos/{task_id}.

POST /v1/videos

cURL

file_to_video

curl --request POST \
  --url https://baze-api.powerbuyin.top/v1/videos \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "wan3.0-video",
  "prompt": "Generate a presentation video based on the file content, with key information presented in animated form.",
  "media": [
    {
      "type": "file",
      "url": "https://example.com/assets/document.pdf"
    }
  ],
  "size": "1080P",
  "ratio": "16:9",
  "seconds": "5"
}
'

200

{
  "id": "<string>",
  "task_id": "<string>",
  "object": "video",
  "model": "<string>",
  "status": "pending",
  "progress": 123,
  "created_at": 123
}

Authorizations

Authorization · string · header · required

Pass Authorization: Bearer <token> in the request header.

Body

application/json

model · enum<string> · required

Model name. Options:

• wan3.0-video: Standard version.
• wan3.0-video-prime: Speed version with identical capabilities but significantly faster end-to-end speed.

Available options: wan3.0-video, wan3.0-video-prime

prompt · string

Text prompt describing the desired video content. Required if media is not provided.

Supports Chinese and English. Each character counts as one, up to 20000 characters (excess will be truncated).

In all-in-one reference mode, use image 1, video 1, etc. in the prompt to reference media items by their order in the media array.

media · object[]

Media asset array supporting images, videos, audio, files, and web links as input. Required if prompt is not provided.

• Each element is a media object with type and url fields.

• In reference-to-video mode, the order in the array defines the reference order in the prompt. Images and videos are counted separately, so image 1 and video 1 can coexist.

• The 1st reference_image corresponds to image 1, the 2nd to image 2, etc.

• The 1st reference_video corresponds to video 1, the 2nd to video 2, etc.

• The 1st reference_audio corresponds to audio 1, the 2nd to audio 2, etc.

Show child attributes

size · enum<string> · default: 1080P

Important: size directly affects cost. Please confirm model pricing before calling.

Video resolution tier. Default is 1080P.

Available options: 480P, 720P, 1080P

ratio · enum<string> · default: adaptive

Video aspect ratio. Default is adaptive (automatically selects an appropriate ratio based on input media and intent).

Options:

• adaptive (default): Adaptive aspect ratio.
• 16:9
• 4:3
• 1:1
• 3:4
• 9:16

Available options: adaptive, 16:9, 4:3, 1:1, 3:4, 9:16

seconds · enum<string> · default: 5

Important: seconds directly affects cost. Please confirm model pricing before calling.

Video duration in seconds. Default is 5.

• Without video input: Integer in range [2, 30].
• With video input: Input video total duration + output video duration must not exceed 30 seconds.
• Pass -1: Smart duration mode, where the model automatically recommends an appropriate duration based on the prompt, content, and rich media.

Available options: -1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30

generate_audio · boolean · default: true

Whether the output video includes audio.

• true (default): Output video includes audio.
• false: Output video has no audio track.

Pricing is the same with or without audio.

prompt_extend · boolean · default: true

Whether to enable prompt intelligent rewriting. When enabled, a large model rewrites the input prompt. Significantly improves results for short prompts, but increases processing time.

• true (default): Enable intelligent rewriting.
• false: Disable intelligent rewriting.

watermark · boolean · default: false

Whether to add a watermark. The watermark is located in the bottom-right corner with the text "AI Generated".

• false (default): No watermark.
• true: Add watermark.

seed · integer

Random seed for reproducibility. Range: [0, 2147483647].

When not specified, a random seed is auto-generated. For better reproducibility, fix the seed value. Note that due to the probabilistic nature of generation, identical results are not guaranteed even with the same seed.

Required range: 0 <= x <= 2147483647

Response

200

application/json

Submission successful, returns a unified video task object.

id · string

Unique task identifier.

task_id · string

Task ID (alias for id).

object · string

Example:

"video"

model · string

Model name used.

status · enum<string>

Task status.

Available options: pending, queued, running, succeeded, failed, cancelled

progress · integer

Task progress percentage.

created_at · integer<int64>

Creation time as Unix timestamp in seconds.