MiniMaxH3
Official MiniMax H3 API · 2K video

MiniMax H3
video generator.

Create up to 2K video from text, opening and closing frames, or multimodal references through the official H3 API.

Try an example
125 credits

Sign in is required before credits are charged.

2K output · 4–15 seconds · Task progress is saved · Download when ready

Text and framesImage, video and audio referencesUp to 12 inputsOfficial H3 workflow

Every model. One studio.

  • Seedance
  • Veo
  • Wan
  • Kling

One model, more context

A multimodal video workflow

MiniMax describes H3 as a general-purpose model that understands text, image, video, and audio inputs in a unified generation workflow.

A runner moving along a misty coastal road at dawn.
01

Text to video

Direct a full scene from a text brief, with prompts up to 7,000 characters.

A lone figure facing a spacecraft in a dark hangar.
02

First and last frames

Use first and last frames or image references to anchor composition and identity.

An unbranded perfume bottle arranged on champagne silk.
03

Multimodal references

Combine up to 12 total reference files, subject to per-media limits. Provide audio references alongside image and video inputs for generation direction.

A passenger train crossing an amber mountain valley at golden hour.2K · 15s
04

Native 2K. Up to 15 seconds.

Maximum output: 2K. Integer duration: 4–15s.

2K

maximum output

4–15s

integer duration

12

mixed references

7,000

prompt characters

SELECTED OUTPUTS

MiniMax H3 Video Examples

Prompts, frames, and references—shown in motion.

01

The Last Fleet

Text to video2K (2560 × 1440) · 24 FPS10.1s

Original prompt

Epic space-opera theatrical teaser: a female captain stands alone before a massive observation window as the last fleet gathers and jumps away in a blinding flash, the bridge shaking, leaving her behind.

From the MiniMax H3 repository

AI-generated example · Source: MiniMax-AI/MiniMax-H3

02

Steam at the Ramen Bar

Image to video2K (2560 × 1440) · 24 FPS8s

Original prompt

Pull focus to the people in the background and add more steam to the ramen bowl.

From the MiniMax H3 repository

AI-generated example · Source: MiniMax-AI/MiniMax-H3

03

Follow the Wind

Video + audio references2K (2560 × 1440) · 24 FPS5.2s

Original prompt

Character speaks: Follow the wind, live free. Leave worries behind, enjoy the moment. Voice timbre follows reference audio 1.

From the MiniMax H3 repository

AI-generated example · Source: MiniMax-AI/MiniMax-H3

More MiniMax H3 video examples ↗

Pricing

Choose the credit option that fits your workflow.

Full-year credits are added at once and remain valid for 365 days.

Explorer

For trying the workflow

$13.25/ month

Billed $159 yearly

19,200credits / year≈ 768s of H3 2K video · about 96 complete 8s clips

Save ≈12% · Failed tasks aren’t charged

Creator

For consistent personal work

$25.75/ month

Billed $309 yearly

39,600credits / year≈ 1,584s of H3 2K video · about 198 complete 8s clips

Save ≈12% · Failed tasks aren’t charged

Recommended

Pro

For serious production

$52.42/ month

Billed $629 yearly

84,000credits / year≈ 3,360s of H3 2K video · about 420 complete 8s clips

Save ≈12% · Failed tasks aren’t charged

Studio

For high-volume creation

$104.92/ month

Billed $1,259 yearly

172,800credits / year≈ 6,912s of H3 2K video · about 864 complete 8s clips

Save ≈12% · Failed tasks aren’t charged

Secure checkout by Stripe · Credits are added after successful payment · Failed provider tasks aren’t charged

A real asynchronous generation flow

How it works

Prompt to video, simply.

  1. 01Describe
    Use a prompt, frames, or multimodal references. The server creates the task.
  2. 02Track progress
    Follow queued and running provider status, with refresh recovery in this browser.
  3. 03Download
    Preview and download only after the provider returns a ready result URL.

The server creates the task. Follow real provider status and recover after refresh in this browser. Download when a result URL is ready. Provider-confirmed failures return reserved credits; uncertain submissions enter reconciliation. Need still images from existing footage? Extract frames from a video. Preview our coming-soon URL-to-video workflow, or for a focused photo-to-video tool see Photo to Video AI.

Start creating  →

Common questions

Questions,
answered.

What to know before you create.

01What can MiniMax H3 generate?

Text-to-video supports 4–15 seconds, 2K, aspect ratio, and an optional visible AIGC watermark. Image mode accepts a first and optional last frame; reference mode accepts supported image, video, and audio reference URLs.

02How are credits calculated?

MiniMax H3 2K generation is quoted at 25 credits per output second. The server-side estimate shown before submission is authoritative.

03When do annual credits arrive?

The annual plan is paid once for the year. The first monthly allocation arrives after successful purchase; the remaining allocations are scheduled monthly, for twelve releases in total.

04Can I export without a watermark?

The generator exposes an AIGC watermark control, so you may turn the visible watermark off. Any legally or provider-required AI labeling still applies.

05Can I use generated videos commercially?

Commercial use is subject to the Terms of Use, rights in every input and reference, provider rules, and applicable law. The service does not clear or transfer third-party rights.

06What happens if a task fails?

Credits reserved for a provider-confirmed create or terminal failure are returned under the existing ledger contract. Ambiguous tasks remain in reconciliation rather than being described as automatically refunded.

Your next frame

Start with
a stronger
prompt.

Describe the scene. Direct the motion. Generate with MiniMax H3.

Start creating Explore prompts

MiniMax H3