01MiniMax H3
Try MiniMax H3, free in ArtCraft.
MiniMax's new multimodal model generates up to 15 seconds of 2K video with native stereo sound. Type a prompt and see for yourself: no credits, no charge.
MiniMax H3Free
0 / 2000
Free MiniMax H3 generations require a free ArtCraft account. New here? You can sign up when you hit Generate.
02Overview
What is MiniMax H3?
MiniMax H3 is the newest generation model from MiniMax, the lab behind the Hailuo video family. Rather than a video-only specialist, H3 is a general-purpose multimodal model: it understands text, images, video, and audio as one unified input, and generates video with native stereo sound in a single pass.
In practice, that shows up as a model that follows instructions unusually well. It renders on-screen text and brand marks accurately, transfers motion from reference footage, keeps characters and style consistent across multi-shot sequences, and produces clips up to 15 seconds at 2K by default, with a faster 768p mode for drafts.
MiniMax is positioning H3 for commercial work (advertising, e-commerce, product design, game content) where control and accuracy matter more than lucky rolls. It is also priced aggressively: MiniMax says 2K output costs less than a third of comparable mainstream models.
On ArtCraft you don't have to take anyone's word for it. H3 generations are free right now: type a prompt at the top of this page and judge the model on your own footage.
03Highlights
Built as one model.
01
Native 2K
Sharp 2K output by default, with a faster 768p mode when you want quick drafts.
02
Up to 15 seconds
Enough room for an actual beat: setup, action, payoff, instead of a four-second loop.
03
Stereo sound built in
Audio is generated together with the picture, so ambience and effects land in sync.
04
Truly multimodal
One model that understands text, images, video, and audio as unified input.
05
Text that reads
Unusually accurate text and logo rendering for signs, packaging, and title cards.
06
Multi-shot scenes
Holds characters and style across cuts for short sequences, not just single shots.
04Examples
See it (and hear it) yourself.
Sample generations from MiniMax's official H3 announcement. The audio is generated with the picture.
- Video: MiniMax
Native stereo sound, generated with the picture 01
- Video: MiniMax
2K performance 02
- Video: MiniMax
Film opening titles 03
- Video: MiniMax
Product website 04
- Video: MiniMax
Animated poster 05
- Video: MiniMax
Advertising and e-commerce 06
05Prompt tips
Get more out of MiniMax H3.
A few habits that pay off with a model that listens this closely.
01
Write the soundtrack too
H3 generates audio with the picture, so tell it what you want to hear: room tone, footsteps, a distant siren, a music cue. Prompts that describe sound get noticeably better mixes.
02
Call the shots like a director
Name the shot type, the camera move, and the action as separate short beats. "Slow dolly-in on..." beats one long run-on wish the model has to untangle.
03
Use real text deliberately
H3 is unusually good at rendering words. If you want a sign, a label, or a title card in the shot, put the exact text in quotes in your prompt.
04
One scene, one idea
Fifteen seconds is a moment, not a movie. Give the clip a single clear action with a beginning and an end, then iterate on the takes you like.
06FAQ
Frequently asked questions.
01Is MiniMax H3 really free on ArtCraft?
Yes. MiniMax H3 generations are currently free for everyone with an ArtCraft account, and creating an account is free too. You can generate right here on this page. If that ever changes, the cost will be shown on the Generate button before you run anything.
02What is MiniMax H3?
H3 is MiniMax's general-purpose multimodal generation model, from the lab behind the Hailuo video family. It takes text, images, video, and audio as unified input and generates video with native stereo sound, with strong instruction following and accurate on-screen text rendering.
03How long and what resolution are the videos?
H3 generates clips up to 15 seconds and is built for 2K output, with a faster 768p mode also available. This page uses the model's default settings to keep things simple; the full ArtCraft app gives you control over resolution, duration, and aspect ratio.
04Does it generate sound?
Yes. H3 generates stereo audio jointly with the video, so ambience, effects, and music land in sync instead of being bolted on afterward. Describe the sound you want in your prompt and it becomes part of the shot.
05Where do my videos go?
Every generation is saved to your ArtCraft library. You can watch and download it right here on this page, or open the ArtCraft app to organize your takes and use them in bigger projects.
06Can I use my own images or footage?
Yes. The full ArtCraft app unlocks the rest of H3: start and end frames, reference images, video, and audio, plus every other leading model in one place.
Every model, one library
One model down. Plenty to go.
What you just used is the real thing: the MiniMax H3 promptbox, settings and all. The ArtCraft app takes it from there. Run the same shot through Seedance, Veo, Sora, Kling, and every other leading model, keep every take in one library, and pick the winner.