Native audio
Direct the sound with the scene
Write dialogue, ambience, or sound effects alongside the visual action. H3 Max generates audio with the video, so this model has no separate audio switch. Listen back before publishing.
H3 Max · Text / Image / Reference
H3 Max AI Video Generator with Native Audio
Turn text or images into short videos with sound. Start with a 5-second H3 Max video at 480P, then refine your next take.
20 welcome credits · 5s · 480P · 16 credits per video
Create a video with H3 Max
2
Ctrl/⌘ + Enter to generate. 0 / 20000
Create a video from a text prompt, without reference uploads.
16
Enter at least 3 characters to continue.
H3 Max samples
Meet your creative Agent
From brief to final cut, with Agent
Describe your idea, review the plan, and refine each shot with Agent. Keep your creative direction in one guided workflow.
A 60-second English walkthrough with narration and captions: brief, Skills, storyboard review, first draft, shot revisions, and export.
Three ways to start
Choose the shot. Shape the sound.
Choose text-to-video, image-to-video or reference-to-video without switching models. Add the inputs your idea needs, then describe what the audience should see and hear.
3 workflows
Start with text, frames or references
Write a prompt, add a first and optional last frame, or combine reference images, videos and audio. In reference mode, use Image 1, Video 1 and Audio 1 to identify your inputs. Balanced prompt expansion stays enabled.
5–15s
Keep each idea focused
Choose any whole-second duration from 5 to 15 seconds. Try a short take first, review its motion and timing, then create a longer version when the direction feels right.
480P / 768P
Frame it for your audience
Choose 480P or 768P. Text-to-video offers six fixed ratios; image-to-video follows the first image. Reference-to-video adds an adaptive option to those six ratios. H3 Max does not offer 2K output here.
Three steps to a first take
How to create a video with H3 Max
You bring the idea. The studio handles the model request, shows its progress, and returns the video for review.
- 1
Write the scene
Describe who or what is in the shot, what moves, and how it sounds. For example: a slow push-in on a ceramic cup, warm window light, rising steam, and quiet café ambience.
- 2
Choose your workflow and inputs
Keep MiniMax H3 Max selected. Choose text, image or reference mode and upload any required media. Set 5–15 seconds and 480P or 768P; image-to-video takes its aspect ratio from the first image.
- 3
Check the quote and generate
Sign in, review the credit cost, and submit your prompt. Check the finished video’s motion, sound, and details before downloading or sharing it.
Studio update · September 4, 2026
What’s new in this studio
See the latest H3 Max workflows and creation settings now available here.
All three H3 Max workflows are live
Start from text, animate a first image with an optional last frame, or combine image, video and audio references—all in one creation surface.
H3 Max is ready by default
New sessions open with H3 Max selected. Choose 5–15 seconds and 480P or 768P, direct native audio in the same prompt, and review the exact credit quote before submitting.
Before you generate
H3 Max FAQ
The essentials on access, video settings, sound, and credits.
What is H3 Max, and how is it different from MiniMax H3?
- H3 Max is a post-trained MiniMax H3 variant tuned for prompt following, visual quality, and faster inference. In this studio it is a separate model with 480P and 768P output; standard MiniMax H3 uses different settings and can include 2K.
Is MiniMax H3 Max available now?
- Yes. This studio connects text-to-video, image-to-video and reference-to-video, with text-to-video selected for new sessions. You need to sign in and have enough credits to generate.
Can I upload images or reference videos to H3 Max here?
- Yes. Image-to-video takes a first image and an optional last frame. Reference-to-video accepts up to 9 images, 3 videos and 3 audio clips, with no more than 12 files in total. Video and audio clips must each be 2–15 seconds, with a 15-second total limit for each media type. Audio must accompany an image or video. Use Image 1, Video 1 and Audio 1 in your prompt to refer to inputs.
Does H3 Max generate sound?
- Yes. H3 Max generates synchronized audio with the video. Describe the dialogue, ambience, or effects you want in the same prompt. This endpoint does not expose an audio on/off control, and the finished soundtrack should be reviewed before use.
What durations, resolutions, and aspect ratios are supported?
- All three workflows support 5–15 seconds in whole-second steps, at 480P or 768P. Text-to-video offers 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16, defaulting to 16:9. Image-to-video follows the first image. Reference-to-video supports those six ratios plus adaptive, its default. New sessions default to 5 seconds and 480P, so the first text or image video fits the 20 welcome credits.
Will my H3 Max video be ready in three seconds?
- Not necessarily. Published performance figures measure backend inference under specific settings. Prompt expansion, queue time, and delivery add to the total wait, so this site does not guarantee a fixed turnaround time.
How much does H3 Max cost here?
- The studio quotes credits before you submit. Text- and image-to-video share duration-based rates and any active launch discount. Reference-to-video uses a separate output rate plus input-token charges beyond a shared 4,096-token allowance; image dimensions and video/audio duration affect that cost. Review the current quote before starting. Credits from other services do not transfer here.
How can I get help with a video task?
- Contact support using the email in the footer and include your task ID so we can locate the generation.
Give your next idea a first take
Start with a scene and a sound. H3 Max is ready in the studio above, with the settings and credit quote visible before you generate.