Create video
Create Wan 3.0 video tasks using the native input structure.
Submit standard-model tasks using model/input/parameters. The native format still uses the platform API address and API Key; no Alibaba Cloud Workspace ID is needed. Asynchronous processing is enabled automatically; clients need not add X-DashScope-Async.
/api/v1/services/aigc/video-generation/video-synthesisAuthorization
bearerAuth Use the platform API Key: Authorization: Bearer your-api-key.
In: header
Request Body
application/json
TypeScript Definitions
Use the request body type in TypeScript.
Submit the native structure. The platform enables asynchronous processing automatically; X-DashScope-Async is not required.
Response Body
application/json
application/json
application/json
Media and combination rules
Provide a prompt or nonempty media. Images, videos, and audio clips are numbered separately by their order within each type in the array; refer to them as “图1”, “视频1”, and “音频1”. No asset reference ID is needed.
| Media | Format and quantity | Key limits |
|---|---|---|
| First/last frames | At most 1 each of first_frame and last_frame | A last frame requires a first frame; cannot be combined with reference images, videos, audio, files, or webpages |
| Reference images | At most 10 reference_image items | JPEG/JPG/PNG (no alpha)/BMP/WEBP; at most 20MB each; each side 240–8000px; aspect ratio at most 8:1 |
| Reference videos | At most 5 reference_video items | MP4/MOV; each 1–15 seconds, total at most 15 seconds; at most 100MB per file; each side 240–4096px; at least 16fps; aspect ratio at most 8:1 |
| Reference audio | At most 5 reference_audio items | WAV/MP3; each 1–15 seconds, total at most 15 seconds; at most 15MB per file |
| File | At most 1 file | At most 100MB; PDF, Word, PowerPoint, Keynote, and Pages at most 50 pages; see the field table for formats |
| Webpage | At most 1 link | Public HTTP/HTTPS webpage, no sign-in; cannot be combined with a file |
First/last frames have the same format, size, and dimension limits as reference images. Images accept public URLs or Base64 with a format prefix, such as data:image/png;base64,...; other media use URLs. Regular media also accept accessible OSS addresses; webpages use HTTP/HTTPS only. Except for webpages, URL extensions must match the media format. Extensionless URLs must still meet the actual content requirements.
Reference images, videos, and audio can be combined. A file or webpage can also accompany these references, but file and link are mutually exclusive. Enable prompt_extend for files or webpages. Use reference videos and prompts for editing and extension; use ratio: "adaptive" for extension.
Duration is -1 (intelligent duration) or an integer from 2～30 seconds. With reference videos, total input and output video duration must not exceed 30 seconds.
Creation results and next steps
After submission, store output.task_id and use native retrieval for status and results. request_id is for tracing, not task queries. Do not recreate the same task to check progress.
For common platform fields, use platform-format creation. Do not mix the two request formats.
Common errors
For HTTP 400, use the error to check JSON, model name, input/parameters, and media combinations. Errors may use code/message/data or a nested error object. When contacting support, provide the error and request ID, not your API Key.