MiniMax H3 - Multimodal AI Video Generator
Create 4-15s AI videos from text, images, first/last frames, and reference images, videos, or audio with MiniMax H3.
Create Your Video
Preview
Ready to Create Your Video
What is MiniMax H3?
MiniMax H3 is a next-generation multimodal AI video model that turns text, images, first/last frames, and reference images, videos, or audio into polished 4–15 second clips. It supports 2K output with adaptive aspect ratio for image-guided workflows, making it ideal for product shots, character motion, stylized scenes, and fast creative iteration.
Why MiniMax H3?
One Model for Text, Image & Reference
Start from a text prompt, animate a single image, guide motion with first and last frames, or attach reference images, videos and audio. H3 adapts the workflow to the assets you already have.
Flexible 4–15 Second Clips
Dial in the exact clip length you need, from quick 4-second loops to fuller 15-second scenes. Per-second pricing keeps short tests affordable and longer shots predictable.
Adaptive & Cinematic Ratios
Choose 16:9, 9:16, 1:1, 4:3, 3:4 or 21:9 for text-to-video. Image and reference modes automatically use adaptive ratio so your source framing stays intact.
How to use MiniMax H3?
From a prompt or your own assets to a finished clip in four steps
Choose a Mode
Pick Text-to-Video, Image-to-Video, First & Last Frame, or Reference Assets mode.
Add Your Input
Write a prompt, upload an image or frames, or attach reference media.
Set Duration & Quality
Choose 4–15s duration and an aspect ratio for text mode. Output is always 2K.
Generate
Click Generate and download your MiniMax H3 video when it is ready.
MiniMax H3 Frequently Asked Questions
Everything you need to know about MiniMax H3 video generation
What can MiniMax H3 generate?
MiniMax H3 generates 4–15 second videos from text prompts, single images, first/last frame pairs, or multimodal reference assets including images, videos and audio.
What resolutions and durations are supported?
H3 generates 2K videos with any whole-second duration from 4 to 15 seconds. Text-to-video supports 16:9, 9:16, 1:1, 4:3, 3:4 and 21:9; image and reference modes use adaptive ratio.
How much does MiniMax H3 cost?
Pricing is per second of output video: 30 gems/s at 2K. A 5s clip costs 150 gems; the cost display updates before you generate.
How many reference assets can I attach?
You can attach up to 9 reference images, 3 reference videos and 3 reference audios, with 12 assets in total. Reference audio must be combined with at least one image or video.