Available now

MiniMax H3 Max: video from references

By MiniMax

Write a scene and turn it into a video of up to 15 seconds. Other modes animate a photo or change a clip you already have.

Model ID
Maker
MiniMax

Model id

Use this in every request
minimax/h3-max/reference-to-video

This model has other modes with their own ids: minimax/h3-max/text-to-video, minimax/h3-max/image-to-video, minimax/h3-max/camera-controls, minimax/h3-max/lip-sync/image-to-video, minimax/h3-max/extend-video, minimax/h3-max/3d-to-video, minimax/h3-max/relight.

Run it

Quote the input, start the generation with the quote, then poll the job. Set RAYWAKE_API_KEY first; see Authentication.

# 1. Price the exact input (quotes are valid for up to 10 minutes)
curl -s https://api.raywake.com/v1/quotes \
  -H "Authorization: Bearer $RAYWAKE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "minimax/h3-max/reference-to-video",
  "input": {
    "prompt": "The woman in Image 1 sits down at the café table from Image 2, pours a cup of tea, then looks up and smiles at the camera.",
    "prompt_expansion_mode": "balanced"
  }
}'

# 2. Start the generation with the quote_id from step 1
curl -s https://api.raywake.com/v1/generate \
  -H "Authorization: Bearer $RAYWAKE_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{"model": "minimax/h3-max/reference-to-video", "quote_id": "QUOTE_ID", "input": {"prompt":"The woman in Image 1 sits down at the café table from Image 2, pours a cup of tea, then looks up and smiles at the camera.","prompt_expansion_mode":"balanced"}}'

# 3. Poll until the status is terminal
curl -s https://api.raywake.com/v1/jobs/JOB_ID \
  -H "Authorization: Bearer $RAYWAKE_API_KEY"

Input

Input fields for minimax/h3-max/reference-to-video
NameTypeDescription
promptrequiredstring
Text prompt for video generation. Refer to reference assets by their modality and order in the reference lists: Image 1, Image 2, Video 1, Audio 1, and so on.
reference_image_urlsarray of image URLs
URLs of subject/style reference images, referenced in the prompt as Image 1, Image 2, and so on. Reference images, videos, and audio clips must add up to at most 12 files.
target_audio_urlstring (URL)
Optional URL of an audio clip at least 2 seconds long (maximum 15 MB) to pin to the generated soundtrack. Cannot be combined with reference audio. Longer clips are trimmed to the generated video's length, keeping the beginning; shorter clips are padded with silence. The clip is resampled to 32 kHz stereo and replaces the output soundtrack without changing playback speed. Accepts an HTTP(S) URL or a base64 data URI.
durationnumber
Video length in seconds. The output can run up to about 0.7 s longer than requested. Use at least 2 s with target_audio_url.

Default: 5

resolutionstring (enum)
The native generation resolution, or 1080P latent refinement from a native 768P source.

Default: 768P

One of: 480P, 768P, 1080P

seedinteger
Random seed. A random seed is selected when omitted.
sync_modeboolean
Return the generated video as base64 instead of a CDN URL.

Default: false

prompt_expansion_moderequiredstring (enum)
How much effort to spend rewriting the prompt before generation. 'disabled' skips prompt expansion. 'balanced' returns in about a second. 'quality' spends up to ~30s on a richer prompt.

Default: balanced

One of: balanced

aspect_ratiostring (enum)
The aspect ratio of the generated video.

Default: adaptive

One of: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16

Output

A succeeded job lists its results under outputs, stored on our side. See Jobs for every field.

Job excerpt
"outputs": [
  { "type": "video", "url": "https://…", "mime_type": "video/mp4", "thumbnail_url": "https://…" }
]

Pricing

  • up to, default options$0.64/ run

The exact charge for your input comes from a quote. How charges settle

API docs

About this model

Overview

MiniMax H3 Max turns a written scene into a video clip of up to 15 seconds at 480P, 768P or 1080P, in six aspect ratios from 21:9 to 9:16. Before it renders, it can rewrite a short prompt into a fuller one; you choose how much effort that takes, or turn it off.

The family goes beyond text: animate a photo, direct the camera, insert something into an existing clip, extend a clip or turn a 3D scene into video. Each mode has its own page and form. You can also pin your own audio file as the soundtrack.

What you can build

  • Ad and social clips from a written brief
  • A moving version of a product photo or an illustration
  • Camera moves you choose, such as a push-in or an orbit, on a still image
  • Longer scenes by extending a clip you already have
  • Clips timed to your own voice-over or music

Prompting tips

  • Turn prompt expansion off when the model should follow your exact words; leave it on for a one-line idea.
  • Describe the camera as part of the scene: where it starts, which way it moves and how fast.
  • When you add your own audio, keep the clip about as long as the audio.
  • To relight a clip, upload an image of a sphere lit the way the scene should look; the result keeps the length of your clip.

What to check

Watch the whole clip and listen to the soundtrack. Check the bicycle wheels, body movement, background continuity and any requested speech. Keep the original input and selected settings when comparing another mode.

Inputs

  • Prompt (required): Text prompt for video generation. Refer to reference assets by their modality and order in the reference lists: Image 1, Image 2, Video 1, Audio 1, and so on.
  • Reference Images (optional): URLs of subject/style reference images, referenced in the prompt as Image 1, Image 2, and so on. Reference images, videos, and audio clips must add up to at most 12 files.
  • Target Audio (optional): Optional URL of an audio clip at least 2 seconds long (maximum 15 MB) to pin to the generated soundtrack. Cannot be combined with reference audio. Longer clips are trimmed to the generated video's length, keeping the beginning; shorter clips are padded with silence. The clip is resampled to 32 kHz stereo and replaces the output soundtrack without changing playback speed. Accepts an HTTP(S) URL or a base64 data URI.
  • Duration (optional): Video length in seconds. The output can run up to about 0.7 s longer than requested. From 0.92 to 15.
  • Resolution (optional): The native generation resolution, or 1080P latent refinement from a native 768P source. Options: 480P, 768P, 1080P.
  • Seed (optional): Random seed. A random seed is selected when omitted.
  • Aspect Ratio (optional): The aspect ratio of the generated video. Options: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16.

Output and limits

  • Length: 0.92 seconds to 15 seconds.
  • Resolutions: 480P, 768P, 1080P.
  • Aspect ratios: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16.
  • Readable text inside the video, especially small text, is unreliable.
  • Scenes with several interacting characters can lose consistency.
  • Longer clips can show more motion drift or scene changes.
  • Low-quality inputs cause glitches: heavy compression, collages, screenshots and text overlays.
  • Some prompts or outputs can be refused by safety checks.

Price guide

  • up to, default options: $0.64 / run