Sign in

MiniMax H3 Max — AI video generator

Try a focused prompt

Enter a prompt before generating.

Preview

Build scenes from image, video and audio references with MiniMax H3 Max

Combine scene images, motion clips and audio references with MiniMax H3 Max. Choose text-to-video, first/last-frame animation or reference-to-video at 480p and 768p. Its reference workflow is useful when a shot must coordinate subject appearance and action. The first two reference images are free upstream; further images and video input add charges.

5–15 s · 480p / 768p
Clip duration
5–15 s

One video clip is returned per generation request.

Output resolution
480p / 768p

Resolution and duration both affect the credit price.

Reference inputs
Up to 12 assets

Up to 9 images, 3 videos and 3 audio files, within 12 total.

Choose the workflow that matches your brief

A focused starting point

Give each reference a single role: cup appearance, pouring motion or café ambience. Keep the action short and inspect object contact and audio timing in the final clip.

Know the boundaries

Generate 5–15 seconds; 2K is not available. Reference mode accepts up to 9 images, 3 videos and 3 audio files, with at most 12 assets total. Audio-only references are not accepted; video and audio are each limited to 15 seconds combined.

Review and download

Watch the full clip before downloading. Check subject continuity, object contact, camera movement and any generated audio. Review the beginning and ending frames before using it in a sequence.

Create with a focused brief

01

Prepare the source and brief

Choose a mode first. Use clear source media you have permission to use, identify each reference’s role, and state what should remain unchanged.

02

Set scope and review the price

Select the smallest useful resolution, duration or image count for the first attempt. Review the displayed credits before submitting. Change one setting at a time when comparing versions.

03

Review and download

Watch the full clip before downloading. Check subject continuity, object contact, camera movement and any generated audio. Review the beginning and ending frames before using it in a sequence.

Questions before you create

What are the important limits?

Generate 5–15 seconds; 2K is not available. Reference mode accepts up to 9 images, 3 videos and 3 audio files, with at most 12 assets total. Audio-only references are not accepted; video and audio are each limited to 15 seconds combined.

How is the displayed price calculated?

The generator displays the site’s credit price for the selected settings. Resolution, duration, output count and reference inputs can affect it. Provider dollar estimates are not the site’s selling price. Review credits before submitting; generating another version is a new paid request.

Create with a focused brief

Give each reference a single role: cup appearance, pouring motion or café ambience. Keep the action short and inspect object contact and audio timing in the final clip.

5–15 s

Clip durationOne video clip is returned per generation request.

480p / 768p

Output resolutionResolution and duration both affect the credit price.

Up to 12 assets

Reference inputsUp to 9 images, 3 videos and 3 audio files, within 12 total.

Check audio cues

TipCompare sound timing with the subject’s action.