Skip to main content
POST
  • Async task API, returns task ID upon submission
  • Supported model: MiniMax-Hailuo-02
  • Three generation modes: text-to-video, image-to-video, first-last-frame generation
  • Supports 512P / 768P / 1080P resolution, 6-second and 10-second duration
Please provide publicly accessible image URLs. Do not pass base64 image data directly in image_urls or metadata; upload local images first using the Upload Image API to get URLs.

Authentication

string
required
All endpoints require Bearer Token authentication.

Request Parameters

string
required
Model name, fixed as MiniMax-Hailuo-02.
string
Video content description, max 2000 characters.
  • Required for text-to-video mode
  • Optional for image-to-video and first-last-frame modes (can describe desired motion)
Supports 15 camera directives (e.g., [Push in], [Pull out], [Pan left, Pedestal up]), see the Camera Directives table below.Example: "A mouse runs toward the camera, smiling and blinking"
integer
default:"6"
Video duration in seconds.Options: 6 or 10
When using 1080P resolution, only 6 seconds is supported.
string
default:"768P"
Video resolution.Options:
  • 512P - Standard definition (available for image-to-video)
  • 768P - HD (default)
  • 1080P - Full HD (6-second duration only)
string[]
First frame image URL array.
  • Not provided: text-to-video
  • 1 image: image-to-video, image used as video starting frame
Image requirements:
  • Format: JPG, JPEG, PNG, WebP
  • Size: under 20MB
  • Dimensions: short side > 300px, aspect ratio between 2:5 and 5:2
Example: ["https://example.com/first-frame.jpg"]
object
Extended parameters.

Three Generation Modes

Resolution and Duration Combinations

Camera Directives

Use [directive] syntax in prompt to control camera movement: Usage rules:
  • Combine up to 3 directives in one bracket for simultaneous movement, e.g., [Pan left, Pedestal up]
  • Sequence directives in the prompt for ordered camera moves, e.g., "...[Push in], then...[Pull out]"

Response

string
Task ID for querying task status.
string
Object type, always generation.task.
string
Model name used for this request.
string
Task status: queued, in_progress, completed, or failed.
integer
Task creation timestamp.

Examples

Text-to-Video

Image-to-Video

First-Last-Frame Generation

Camera Control

10-Second Long Video + Fast Preprocessing

Video generation is an async task. Use Get Video Task Status to query progress and results.