New model — live now

MiniMax H3 Video Generator — 2K Video With Stereo Sound

Turn a prompt, a photo, or reference images into 2K video with audio already in place. Renders 4 to 15 second clips with no editing timeline.

2K OutputStereo Audio4–15 Seconds

Dialogue, Music, and Effects Land With the Picture

Every render comes out at 2K with native stereo audio — spoken lines, ambience, and music in the same pass. What you download is what you post.

Upload a first frame and add a last frame to build motion between the two. Orientation follows your image automatically.

Feed up to 9 reference images plus short video and audio clips. Hold your subject look from shot to shot.

Handles kinetic type, manga-panel energy, and logo-heavy edits — holds the design language for the full clip.

Capabilities

Dialogue, ambience, effects, and music render together with pixels — no silent draft to fix later.

Text-to-video, image-to-video, and reference-to-video in one model. Clips from 4 to 15 seconds.

Upload first and last frames to define motion between two stills — portraits stay portraits, wide shots stay wide.

First 5 references are free. Multiple angles of the same character give much more to hold onto.

Chases, impacts, slow-motion beats — write the shot list and MiniMax H3 follows with camera moves and SFX.

A 6-second video costs about 55 credits. Failed generations are refunded automatically.

Write a prompt, upload a first frame to animate a still, or add reference images for consistency.

Choose 4 to 15 seconds and the aspect ratio your channel needs. Audio is part of every render.

MiniMax H3 returns a 2K clip with stereo sound in a few minutes. Keep it, tweak, or post as is.

Prompt, photo, or reference stack — 2K clip with sound in a couple of minutes.