Jimeng video generation

Jimeng video submits async tasks through an OpenAI-compatible endpoint, primarily via multipart/form-data.

POST /v1/videos

Pass a single reference image file with . A successful submission returns id and status. Poll results with Task query.

Available models

Headers

Name	Type	Required	Description
	string	Yes	Request authentication. Use a Bearer Token, for example:

This endpoint uses multipart/form-data (-F sets Content-Type automatically). Upload the reference image with -F input_reference=@....

Request body (form fields)

Name	Type	Required	Description
	string	Yes	Model name
	string	Yes	Prompt. Supports Chinese and English
	file	No	Reference image file (image-to-video). Supports png / jpeg / jpg / webp
	string	No	Duration in seconds: , , , default
	string	No	Size: (portrait, default) or (landscape)

Request example

Text-to-video

curl -X POST https://octopusx.ai/v1/videos \
  -H "Authorization: Bearer <token>" \
  -F "model=jimeng-video-3.0" \
  -F "prompt=A cat chasing butterflies on the grass" \
  -F "seconds=5" \
  -F "size=720x1280"

Image-to-video

curl -X POST https://octopusx.ai/v1/videos \
  -H "Authorization: Bearer <token>" \
  -F "model=jimeng-video-3.0" \
  -F "prompt=Make the person walk forward and smile" \
  -F "[email protected]" \
  -F "seconds=5"

Response example

{
  "id": "video_123",
  "task_id": "video_123",
  "object": "video",
  "model": "jimeng-video-3.0",
  "status": "queued",
  "progress": 0,
  "created_at": 1712697600
}

Response fields

Name	Type	Description
	string	Task ID for subsequent polling
	string	Task ID (compatibility field; same value as )
	string	Always
	string	Model actually used
	string	Task status: , , , , and others
	integer	Progress percentage
	integer	Creation timestamp

Current rules

Item	Rule
Default duration	Defaults to seconds when not explicitly set
Default size	Defaults to when not explicitly set
Reference image	Upload a single image file via
Reference image formats	Supports png / jpeg / jpg / webp

Other formats and modes

Doubao channel

Also submitted via POST /v1/videos (multipart). Models and fields depend on what is currently available for that channel.

Related endpoints

• Video APIs overview
• Jimeng task query
• Error codes