Skip to main content
POST
Wan 3.0 All-in-One Video Generation
Note for users in mainland China: Please use https://toapis.cn as the API endpoint (Base URL). Replace https://toapis.com with https://toapis.cn in the examples in this document.
The public model name for Wan 3.0 is wan3.0-video. A single model supports the following generation modes:
  • Text-to-video
  • First-frame or first/last-frame control
  • Joint generation with reference images, reference videos, and reference audio
  • File or public web page references
The API is asynchronous. A successful submission returns a task ID; use the Video Task Status endpoint to query progress and results.
First/last-frame mode and reference-asset mode are mutually exclusive.If the request contains first_frame or last_frame, you cannot also pass reference images, reference videos, reference audio, files, or web links. Reference images, reference videos, and reference audio can be combined with each other.

Authentication

string
required
Use Bearer Token authentication:

Request Parameters

string
default:"wan3.0-video"
required
Model name, fixed as wan3.0-video.
string
Description of the video content, subject motion, camera work, and sound. At least one of prompt or media assets must be provided.When using multimodal references, you can write “image 1”, “video 1”, “audio 1” in the prompt, matching the array order.
integer
default:"5"
Output duration, supports 2 to 30 seconds. Pass -1 for smart duration to let the model decide the output length automatically.
string
default:"adaptive"
Output aspect ratio, supports:
  • adaptive, adapts to reference assets
  • 16:9
  • 9:16
  • 1:1
  • 4:3
  • 3:4
The compatible field aspect_ratio can also be used. Do not pass both fields.
string
default:"1080p"
Output resolution, supports 480p, 720p, and 1080p.
boolean
default:true
Whether to generate audio. Enabled by default.
boolean
default:false
Whether to add a watermark. Disabled by default.
integer
Optional random seed. The same parameters and seed help reproduce similar results.
object[]
First/last-frame control array. Each item contains:
  • url: publicly accessible image URL
  • role: first_frame or last_frame
Provide at most one first frame and one last frame. last_frame requires first_frame to be provided as well.
string[]
Array of reference image URLs, up to 10 images. Cannot be mixed with first/last frames in image_with_roles.
object[]
Array of reference videos, up to 5 clips with a total duration not exceeding 15 seconds. Each item uses the following structure:
object[]
Array of reference audio, up to 5 clips with a total duration not exceeding 15 seconds. Each item contains:
  • url: publicly accessible audio URL
  • role: fixed as reference_audio
object
File or web page references are passed via metadata.media.
Local images or videos must be uploaded first to obtain a public URL. Do not pass local file paths or base64 data directly in the generation request.

Asset Combination Rules

If you receive the following error:
Remove the first/last frames and keep the reference assets; or remove all reference assets and keep only the first/last frames.

Request Examples

Text-to-Video

First/Last-Frame Video

Reference-Video Generation

Do not pass first_frame or last_frame in this request.

Combined Image, Video, and Audio Reference

File Reference

Response

string
Video task ID, used to query task status.
string
Object type, usually generation.task.
string
Model name used for this request.
string
Task status: queued, in_progress, completed, or failed.
integer
Task progress, ranging from 0 to 100.
integer
Task creation timestamp.