Skip to main content
SeeVideoSeeVideo

MiniMax H3 Video Generator

Start with a focused scene, choose one of the supported output shapes, and review the returned video before publishing it.

Resolution
768p
Sign in to generate
Optional image references
Output preview

Your generated video will appear here when processing is complete.

Your generated video will appear here when processing is complete.

What MiniMax H3 can do in this workspace

MiniMax H3 is a short-form video model for turning a scene description into a moving clip. This page currently provides text prompts, optional image references, 4–15 second outputs, three aspect ratios, and 768p video.

How to create a video with MiniMax H3

A clear input makes a short generation easier to review. Keep the first attempt narrow, then change one variable at a time.

  1. 01

    Write a focused scene prompt

    Name the subject, action, location, lighting, camera movement, and intended mood. Avoid packing unrelated scenes into one short clip.

  2. 02

    Add image references when needed

    Use an image reference when appearance, palette, or framing matters. Give each image a clear job and confirm that you have permission to use it.

  3. 03

    Choose settings and review

    Select 4–15 seconds and a supported ratio, submit the task, then check motion, identity, text, sound, and rights before publishing.

MiniMax H3 video settings and limits

These are the video settings currently available for MiniMax H3 on this page.

SettingAvailable valueNotes
Model IDminimax-h3The model used for this workspace.
InputsPrompt plus optional image referencesUp to 9 reference images, plus optional first and last frame images.
Duration4–15 secondsInteger seconds only.
Aspect ratio16:9, 9:16, or 1:1Choose the format that matches the final placement.
Resolution768pThe available output resolution on this page.
OutputVideo outputAvailable after processing is complete.

Prompt patterns for MiniMax H3

Build a clear prompt by stating what should move, what should stay stable, and where the camera is placed.

Product movement

16:9 · 5 seconds · 768p

A matte black travel bottle on a pale stone table. A slow camera arc from left to right, one soft highlight moving across the cap, clean studio background, stable product shape, no added text or logos.

Useful for checking shape stability, camera pacing, and unwanted details.

Vertical atmosphere

9:16 · 5 seconds · 768p

A cyclist pauses beside a quiet city canal at blue hour, then looks toward lights reflected in the water. Vertical composition, gentle handheld motion, natural clothing movement, restrained color grade.

Keep the subject and one action central when the frame is tall.

Reference-led motion

1:1 · 6 seconds · one image reference

Use the reference image for the subject's colors and silhouette. The subject turns slowly toward camera while the background remains calm and softly lit. Preserve the main shape; no new people, signs, or captions.

Use the reference to define composition, subject, or style, then describe the motion you want.

MiniMax H3 pricing and credits

Review the current pricing page for available plans and credit terms before generating a production video.

View pricing

MiniMax H3 FAQ

What is MiniMax H3 in this page?

MiniMax H3 is the model selected for this video workspace. The settings on this page show the inputs and output options currently available here.

Can MiniMax H3 generate video from text?

Yes. Start with a text prompt and add image references when they help establish appearance or framing. Keep the scene focused for a short output.

Which duration and resolution settings are available?

This page accepts integer durations from 4 to 15 seconds, uses 768p output, and provides 16:9, 9:16, and 1:1 ratios.

How are MiniMax H3 credits calculated?

This page does not state a model-specific credit price. Review the current Pricing page for applicable site terms before generating.

What happens when generation fails?

If generation fails, no output is returned. Review the prompt and settings, then try again with a simpler scene.