Start free

MiniMax H3 API

Generate video from a prompt with native audio in stereo, at resolutions up to 4K.

Start Free Docs

MiniMax H3 modes

MiniMax · minimax-h3-text-to-video

Text to video

Generate video from a prompt with native audio in stereo, at resolutions up to 4K.

Put a prompt and "model": "minimax-h3-text-to-video" on a video clip, and the render generates it and edits it into the video. To generate it on its own, send the same asset to the generate endpoint.

Built for4K clips with stereo sound · Product and brand scenes · Repeatable variations

MiniMax H3 “Soft light sweeps across a pair of white headphones on a blush-pink backdrop”
Technical specifications
Input Text prompt
Output Video
Length 5 to 15 seconds
Resolution 480p, 768p, 2k, 4k
Aspect ratios 21:9, 16:9, 4:3, 1:1, 3:4, 9:16
Audio Native audio
Seed Set a seed for repeatable results
Prompt enhancement On by default, can be turned off
JSON
{
  "asset": {
    "type": "video",
    "prompt": "Soft light sweeps across a pair of white headphones on a blush-pink backdrop",
    "model": "minimax-h3-text-to-video",
    "options": {
      "aspectRatio": "4:3",
      "resolution": "768p"
    }
  },
  "length": 5
}
Example request to the generate endpoint.

Options

Option Values Default
aspectRatio 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 16:9
resolution 480p, 768p, 2k, 4k 768p
enhancePrompt boolean true
seed integer

MiniMax · minimax-h3-image-to-video

Image to video

Animate an image with native audio in stereo. Supply an ending image to guide the final frame.

Put a prompt and "model": "minimax-h3-image-to-video" on a video clip, and the render generates it and edits it into the video. To generate it on its own, send the same asset to the generate endpoint.

Built forAnimating a still with stereo sound · Endings set by an end image · Product moments up to 4K

The image the example starts from
Start image
MiniMax H3 “The dried flowers sway gently as the camera pushes in”
Technical specifications
Input Text prompt and a start image, with an optional end image
Output Video
Length 5 to 15 seconds
Resolution 480p, 768p, 2k, 4k
Audio Native audio
Seed Set a seed for repeatable results
Prompt enhancement On by default, can be turned off
JSON
{
  "asset": {
    "type": "video",
    "prompt": "The dried flowers sway gently as the camera pushes in",
    "model": "minimax-h3-image-to-video",
    "options": {
      "startSrc": "https://shotstack.io/assets/generated/models/nano-banana-pro.webp",
      "resolution": "768p"
    }
  },
  "length": 5
}
Example request to the generate endpoint.

Options

Option Values Default
startSrc (required) string
endSrc string
resolution 480p, 768p, 2k, 4k 768p
enhancePrompt boolean true
seed integer

Pricing

Generation is paid in credits from the same balance as rendering, at what the model's provider charges plus a small margin. See AI generation pricing.

Mode Charged per Price changes with Length a quote assumes without one
Text to video Second of video, 5 to 15 Resolution: 480p, 768p (default), 2K, 4K 5 seconds
Image to video Second of video, 5 to 15 Resolution: 480p, 768p (default), 2K, 4K 5 seconds

Get a quote

POST /generate/quote takes the same body as the generate endpoint and returns what that generation would cost. Nothing is generated, reserved or charged.

Bash
curl -X POST \
     -H "Content-Type: application/json" \
     -H "x-api-key: $SHOTSTACK_API_KEY" \
     -d '{
       "asset": {
         "type": "video",
         "prompt": "Soft light sweeps across a pair of white headphones on a blush-pink backdrop",
         "model": "minimax-h3-text-to-video",
         "options": {
           "aspectRatio": "4:3",
           "resolution": "768p"
         }
       },
       "length": 5
     }' \
     https://api.shotstack.io/edit/stage/generate/quote

The response gives credits, the price of that request, and ceiling, which is true when the request has no length for a model charged by length, so the price is an estimate. Prices follow the providers, so ask again rather than storing one.

How it compares

MiniMax H3 beside the other models in Shotstack that make the same thing.

ModelBest forModesLengthMax resolutionAspect ratiosAudioEnd imageSeed
MiniMax H3Clips up to 15 seconds and 4K, with native stereo audioText to video, Image to video5–15 s4K6NativeYesYes
Seedance 2.0Cinematic motion and prompt-directed camera movesText to video, Image to video4–15 s1080p6OptionalYes—
Seedance 2.5Longer scenes, up to 30 seconds in one clipText to video, Image to video4–30 s1080p6OptionalYes—
Wan 3.0 Prime1080p clips, with audio and prompt enhancement on by defaultText to video, Image to video2–30 s1080p5On by defaultYesYes
Gemini Omni Flash 1.1Short clips up to 4K, with native audioText to video, Image to video3–10 s4K2NativeYes—

Questions

Do I need an account with the model provider?
No. Generation runs on your Shotstack API key and is paid in credits from the same balance as rendering, so there is no separate account to open and nothing extra to integrate.
Do I pay again for the same asset?
No. Every generated file is stored against your account. An asset with the same type, model, prompt, options and length as one generated before reuses the stored file at no charge.
What happens if a generation fails?
A failed generation isn't charged. If a render fails after some of its assets were generated, those assets are kept, and rendering again reuses them at no charge.
Can I generate an asset without rendering a video?
Yes. POST the asset to the generate endpoint and you get back a URL to the file, to check it before you render, to reuse it across many edits, or to use it outside Shotstack.

Start free

  • 10 free credits to start
  • No credit card required
Start Free All AI models