Cinematic
A single take through a collapsing seascape
Open the video to review motion, timing, composition, and synchronized sound.
Post-trained H3 variant
MiniMax H3 Max AI Video Generator
A post-trained H3 variant built for stronger prompt adherence, better aesthetics, and ultra-fast video generation with synchronized audio.
Reference Media
Max 12 total: 9 images, 3 videos, 3 audio. Media 2-15s. Files 10/50/10MB.
Prompt *Describe your vision
181/7000
5s10s15s
Available credits: 0
One continuous take, no cuts
Camera direction, movement, and synchronized sound
MiniMax H3 Max
Reference workflowMulti-Reference GuideView guideClose guide
MiniMax H3 Max accepts text, image, video, and audio references. Combine only the materials your scene needs for more precise creative direction.
Prompt Guide
Name uploaded files directly in your prompt.
Use references such as @Image 1, @Video 1, or @Audio 1 so the model knows which material to use.
Material Limits
- Up to 9 reference images
- Up to 3 reference videos, 15 seconds total
- Up to 3 reference audio files, 15 seconds total
- Up to 12 materials across all reference types
Supported Input Combinations
TextText + ImageText + VideoText + AudioText + Image + VideoText + Image + AudioText + Video + AudioText + Image + Video + Audio
Note: Reference images, videos, and audio are optional. Add only the materials needed for your prompt.
Supported file requirements
Images
JPG, JPEG, PNG, WEBP, BMP, TIFF, or GIF. Up to 10 MB each.
Videos
MP4 or MOV. 2 to 15 seconds, up to 50 MB each.
Audio
MP3 or WAV. 2 to 15 seconds, up to 10 MB each.
Five to fifteen seconds
480P, 768P, or 1080P at 24 FPS
Synchronized audio in one pass
How Fast Is MiniMax H3 Max?
About 3 seconds to generate a 5-second 768P video in the developer’s published benchmark.
2.78 seconds
H3 Max generation time
For the published 5-second, 768P example.
53.20 seconds
Standard H3 generation time
The baseline reported alongside the same example.
About 19.1×
Generation speed relative to H3
53.20 ÷ 2.78; approximately 94.8% less generation time.
These are developer-reported benchmark timings, not a MiniMaxm speed test or a per-request guarantee. The separate 35× throughput claim uses a different measurement and is not used in the calculation above. Your total wait can include uploads, queueing, generation, and delivery.
See What MiniMax H3 Max Can Create
Six demonstrations cover cinematic direction, gameplay, dialogue, stop motion, nature, and product imagery.
Retro Gameplay
A side-scrolling city run
Open the video to review motion, timing, composition, and synchronized sound.
Spoken Scene
A direct-to-camera kitchen line
Open the video to review motion, timing, composition, and synchronized sound.
Clay Animation
A scarfed frog serves pancakes
Open the video to review motion, timing, composition, and synchronized sound.
Nature Macro
A hummingbird pauses at a flower
Open the video to review motion, timing, composition, and synchronized sound.
Product Close-up
Espresso forming in macro detail
Open the video to review motion, timing, composition, and synchronized sound.
What Is MiniMax H3 Max?
MiniMax H3 Max is a post-trained variant built from the open-weight MiniMax H3 base model.
MiniMax developed the H3 base model. fal Research developed H3 Max through additional training focused on prompt adherence and visual quality, paired with an optimized inference system. MiniMaxm operates this independent hosted interface.
The hosted settings shown here are designed for rapid text-to-video, image-to-video, and reference-to-video work at 480P, 768P, or 1080P. Standard H3 remains the option for 2K output.
Sources and verification: Model provenance and benchmark figures were checked against the developer's H3 Max launch report, published August 26, 2026. Benchmark figures describe the developer's infrastructure; output controls describe this website. Last checked September 18, 2026.
MiniMax H3 Max Key Features
Each capability is paired with a demonstration showing the behavior in a finished clip.
Camera and Motion
Direct the camera as precisely as the action
A low tracking shot follows a scooter down a steep street while keeping the rider framed through the turn.
First and Last Frame
Connect two stills with continuous movement
Start with one image and define the final frame. The model creates an uninterrupted path between them.
Character Consistency
Keep a character recognizable across cuts
Keep hair, clothing, facial features, and proportions coherent while time, setting, and lighting change.
Native Audio
Create the soundtrack in the same generation
Generate dialogue, ambience, foley, music, and environmental sound alongside the image.
Visual Aesthetics
Better aesthetics through post-training
Post-training targets more appealing composition, lighting, and visual style. Inspect the palette, texture, and framing in this demonstration; the developer reports improved aesthetic preference, while individual results vary.
Prompt Adherence
Stronger adherence to your creative brief
Additional training targets closer alignment with the requested subject, action order, camera direction, and visual details. Check whether each requested beat appears in the clip, rather than judging adherence from a single attractive frame.
MiniMax H3 Max vs MiniMax H3
H3 Max emphasizes fast 768P iteration, while standard H3 supports broader tools and higher resolution.
CapabilityH3 MaxStandard H3
Best fitSpeed and prompt adherenceResolution and broader control
Resolution480P, 768P, or 1080PUp to 2K
Text and image to videoYesYes
Reference to videoYesYes
Video editingLimitedYes
AudioSynchronizedSynchronized
How to Use MiniMax H3 Max
Four decisions take you from a first prompt to a reviewable clip.
Pick text, image, or reference input
Start with text, upload opening and closing frames, or add reference media.
Write the shot in order
List subject, action, camera movement, lighting, and sound in playback order.
Choose output settings
Select 480P, 768P, or 1080P and choose a duration from five to fifteen seconds.
Fix one issue per rerun
Review camera path, event order, consistency, text, and audio sync before refining.
MiniMax H3 Max FAQ
Answers about the model, output limits, speed, audio, and differences from standard H3.
What is MiniMax H3 Max?+
H3 Max is a post-trained variant of MiniMax H3 developed by fal Research. Its post-training targets stronger prompt adherence and better visual aesthetics, while inference optimization enables faster generation with synchronized audio.
Is H3 Max an official MiniMax model?+
MiniMax developed the H3 base model; fal Research developed the post-trained H3 Max variant. MiniMaxm provides an independent hosted interface for using it.
What resolution and duration are supported?+
H3 Max supports 480P, 768P, or 1080P at 24fps, with clip lengths from five to fifteen seconds.
How many credits does 1080P cost?+
H3 Max 1080P output costs 8 credits per second: 40 credits for 5 seconds, 80 for 10 seconds, or 120 for 15 seconds. Reference inputs may add charges; check the generator estimate before submitting.
Does H3 Max support 2K?+
No. Standard MiniMax H3 is the appropriate option when the workflow requires 2K output.
Can H3 Max animate an image?+
Yes. Upload an opening image, describe the movement, and optionally define a final frame.
Can H3 Max use reference media?+
Yes. A request can include images, videos, and audio references together with a prompt.
Does H3 Max generate audio?+
Yes. Dialogue, ambience, foley, music, and effects can be generated in sync with the video.
How fast is H3 Max?+
The developer reports a 5-second 768P clip generated in 2.78 seconds, or about 3 seconds. This is a published benchmark, not a measured MiniMaxm turnaround time. Uploads, queueing, input complexity, and delivery can increase the time you wait.
How much faster is H3 Max than MiniMax H3?+
In the developer's published 768P, 5-second example, standard H3 took 53.20 seconds and H3 Max took 2.78 seconds. Dividing 53.20 by 2.78 gives approximately 19.1 times the generation speed, or about 94.8% less time. This example is distinct from the separately reported 35x throughput claim and is not a guarantee for every request.
H3 Max or standard H3?+
Use H3 Max for fast 768P iteration and prompt adherence. Use standard H3 for 2K output and broader controls.
Make Your First MiniMax H3 Max Video
Start from text or an image and create a 768P clip with synchronized sound.
Generate MiniMax H3 Max Video