Workspace ID is not set. Wan 3.0 uses the endpoint
https://{WorkspaceId}.{region}.maas.aliyuncs.com. Open âš™ Settings.Model
Pro: highest quality, 4K for text-to-image. Standard: faster, up to 2K.
Model
Mode
Prompt 0 / 20000
Refer to inputs by order:
Image 1, Video 1, Audio 1. Wan 3.0 has no negative prompt.With input images, the prompt is the edit/composition instruction; refer to
image 1, image 2…Input images (optional)
Up to 9 images (JPG/PNG/WEBP/BMP, 240–8000px, ≤20MB). No images = text-to-image. bbox (interactive edit):
x1,y1,x2,y2 in input-image pixels, up to 2 boxes separated by ;.Parameters
4K: wan2.7-image-pro, text-to-image, non-sequential only. Editing & image sets max 2K.
The API has no ratio parameter, so this is sent as
size = W*H with the same pixel count as the chosen tier. auto = square for text-to-image, or follows the input image when editing.1–4 (1–12 with enable_sequential).
thinking_mode
Extra reasoning for quality (slower). Text-to-image, non-sequential only.
enable_sequential
Coherent image set (storyboard), n up to 12.
watermark
Add an "AI generated" watermark.
3–10 colors, ratios must total 100%. Not used in sequential mode.
Advanced: extra parameters (JSON, merged into parameters)
Media input
Parameters
adaptive = follow the input media aspect ratio (default).Smart duration (-1)
The model picks the length.
2–30 seconds. With video input: input + output duration ≤ 30 seconds.
audio
Generate dialogue, music & SFX. Off = silent video (same price).
prompt_extend
Auto-expand the prompt. Turn off if your prompt is already detailed (faster).
watermark
Add an "AI generated" watermark.
Advanced: extra parameters (JSON, merged into parameters)
Request preview
Result
Results appear here.
Videos usually take 1–5 minutes, images 10–60 seconds.
Videos usually take 1–5 minutes, images 10–60 seconds.