# MiniMax H3

`minimax/h3` · MiniMax · video

Generate a 4–15 second clip with sound from text, opening and closing frames, or image, video and audio references. The standard H3 tier offers 768p and 2K output.

[Model page](https://sociaro.com/models/minimax/h3) · [JSON](https://sociaro.com/models/minimax/h3.json) · [Model pricing](https://sociaro.com/pricing?model=minimax%2Fh3) · [Documentation](https://docs.sociaro.com/models/minimax-h3/) · [Playground](https://my.sociaro.com/playground/?model=minimax/h3)

Video capabilities — `video-generation`, `video-edit`

## Use cases

- Making a 2K shot guided by visual references
- Directing motion between two storyboard frames

## Technical details

- API model ID — minimax/h3
- Provider — MiniMax
- Modality — Video
- Input — Text prompt · Images · Video · Audio
- Output — Video
- Endpoint — POST /v2/jobs
- Workflow — Omni video
- Resolution — 768P · 2K
- Aspect ratios — adaptive · 21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16
- Duration — 4–15 seconds

## Request parameters

- `content` (array, required) — Describe the clip in one text prompt. To guide the result, add opening or closing frames, or a set of reference images, clips and sounds.
  - Allowed values — `text`, `image_url`, `video_url`, `audio_url`
  - At least 1 item
  - The prompt can contain up to 7,000 characters. Supply media as public HTTPS URLs or data URIs.
  - Choose one input mode per request. Frame mode accepts one opening image, one closing image or both.
  - Reference mode accepts up to 9 images, 3 video clips and 3 audio clips. Each clip lasts 2–15 seconds, with 15 seconds total per media type.
- `resolution` (string, required) — Output resolution of the clip, each tier with its own price per second; this model renders at 768P or 2K.
  - Allowed values — `768P`, `2K`
- `duration` (integer, required) — Length of the clip, 4 to 15 seconds, as a whole number; the output is priced per second, and reference videos and input images beyond the free allowance are billed on top.
  - 4–15
- `ratio` (string) — Default adaptive. A text-only request needs a concrete ratio (not adaptive); with a first or last frame the ratio follows the picture.
  - Allowed values — `adaptive`, `21:9`, `16:9`, `4:3`, `1:1`, `3:4`, `9:16`
- `response_format` (string) — The shape of the generator's answer. Only task, the default, is accepted; the gateway reads the answer in that shape.
  - Allowed values — `task`

## Pricing

Rates in USD. Catalogue timestamp 2026-11-21T02:23:44Z.

- USD 0.096 per second of output video · 768p · active
- USD 0.156 per second of output video · 2K · active
- USD 0.096 per second of reference video · 768p · active · the seconds the generator reports, on the row for the output's resolution
- USD 0.156 per second of reference video · 2K · active · the seconds the generator reports, on the row for the output's resolution
- USD 0.048 per input image after the first 5 · active · first and last frames and reference images all count

- Billed on the seconds and resolution the generator reports for the finished clip.
- Lengths of 4 to 15 s; resolution ("768P" or "2K") and length must be sent.
- Reference audio is free.

## Examples

### Glass feather

A sliver of golden light travels along the ribs of a blue glass feather.

[Result (video)](https://previews.sociaro.com/shared/promo/gallery/h3-leda/sapphire-feather.mp4) · [Poster](https://previews.sociaro.com/shared/promo/gallery/h3-leda/sapphire-feather.webp)
