MiniMax H3: 2K AI Video Generation with Native Audio
Use MiniMax H3 with text, images, keyframes, video, and audio references to guide subject, motion, camera, timing, and sound in one multimodal workflow.
Try it free in most regions worldwide. No credit card required.
MODEL OVERVIEW
What Is MiniMax H3?
MiniMax H3 is a multimodal AI video model that uses text, images, video, and audio as context to generate video with synchronized sound in a single workflow. It can use reference media to guide appearance, motion, camera behavior, pacing, and audio.
Open the MiniMax H3 AI Video Generator →Multimodal Video Generation
MiniMax H3 uses text, images, video, and audio in one shared context, letting you guide the subject, motion, camera, and sound in a single generation.
Shot-Level Creative Control
Direct camera movement, dialogue, pacing, ambience, and sound cues in the same prompt, giving each generation a clearer visual and audio brief.
Reference & Keyframe Control
Use images to guide appearance, video references to influence motion, audio to shape timing, or first-and-last frames to control how a shot begins and ends.
Text Prompt
Guide the subject, action, camera, lighting, dialogue, and sound.
Image Reference
Guide appearance, product details, style, and composition.
Video Reference
Guide motion, performance, camera movement, and timing.
Audio Reference
Guide rhythm, dialogue, ambience, music, and sound cues.
MiniMax H3 Video Examples
See real prompts, generation settings, and outputs across product shots, character scenes, cinematic motion, and reference-led workflows.
Browse the MiniMax H3 Prompt Guide →Prompt
A luxury stainless-steel wristwatch rotates slowly on a matte black pedestal in a dark studio. A narrow strip light sweeps across the polished metal and glass as the camera makes a smooth macro push-in. Keep the watch shape and proportions consistent throughout the shot. Premium commercial lighting, deep black background, crisp reflections, subtle mechanical ticking and quiet studio ambience. No text, logos, or extra objects.
Prompt
A black sports car drives through a rain-soaked downtown street at night while the camera tracks smoothly beside it at a low angle. Red and blue neon signs reflect across the wet pavement and glossy bodywork. Keep the car geometry and wheel motion realistic and consistent. Cinematic contrast, light rain, subtle atmospheric haze, engine sound, wet-road tire noise, and distant city ambience.
Prompt
A woman stands beside a large apartment window in warm late-afternoon light. She slowly turns her head toward the camera while the camera makes a gentle handheld push-in. Keep her face, hairstyle, clothing, and body proportions consistent. Natural skin texture, shallow depth of field, soft shadows, subtle fabric movement, quiet room ambience, and distant city traffic.
Prompt
A clear glass perfume bottle stands centered on a glossy black surface as soft red light gradually reveals its edges. Fine mist drifts slowly behind the bottle while the camera pushes forward toward a clean hero shot. Keep the bottle perfectly stable with realistic glass reflections and refraction. Luxury advertising style, dark background, controlled highlights, subtle atmospheric sound, no text or logos.
Prompt
Morning sunlight enters a minimalist concrete interior through tall windows as the camera glides slowly forward through the space. Soft shadows move across the walls and floor, revealing the texture of concrete, wood, and glass. Keep architectural lines straight and stable. Calm cinematic composition, natural daylight, subtle room tone, distant outdoor ambience, and no people.
Prompt
A lone explorer walks slowly through a vast futuristic industrial chamber illuminated by cool white floor lights and distant red warning lights. The camera follows from behind with a steady cinematic tracking shot. Steam drifts from machinery while small particles float through the air. Keep the character and environment consistent. Deep mechanical ambience, soft footsteps, distant machinery, and restrained cinematic tension.
Four Ways to Create with MiniMax H3
Choose a workflow based on what you already have: a text prompt, one image, two keyframes, or a set of multimodal references.
Text to Video
Start with a prompt describing the subject, action, camera, lighting, style, and sound. MiniMax H3 generates a complete video at up to 2K in six aspect ratios from 21:9 to 9:16.
Generate Text to VideoImage to Video
Upload a product photo, illustration, or storyboard frame to define the starting look, then guide motion, camera movement, timing, atmosphere, and sound with your prompt.
Animate an ImageFirst & Last Frame
Set the opening and ending frames, then let MiniMax H3 create the transition between them for more controlled reveals, transformations, and scene changes.
Generate from First & Last Frames
Sound input
Reference to Video
Combine image, video, and audio references to guide appearance, motion, camera behavior, timing, and sound. Use this workflow when a text prompt alone is not enough to define the result.
Up to 9 images · 3 videos · 3 audio files
Generate from ReferencesHow to Create a Video with MiniMax H3
- 01
Add References
Use images, video, or audio to guide appearance, motion, timing, and sound.
- 02
Describe Your Shot
Define the subject, action, camera, style, and sound.
- 03
Generate & Refine
Generate your video, review the result, and adjust one instruction at a time.
MiniMax H3 vs H3 Max
Compare the standard H3 workflow with the speed-focused H3 Max experience.
| Capability | MiniMax H3 | MiniMax H3 Max |
|---|---|---|
| Focus | Multimodal control and native audio | Faster generation and iteration |
| Inputs | Text, images, video, and audio references | Text, images, keyframes, video, and audio references |
| Duration | Up to 15 sec | Up to 15 sec |
| Resolution | Up to 2K | 480p and 768p |
| Audio | Native stereo audio | Audio generated with the video |
| References | Rich multimodal reference workflow | Image, video, and audio references; unavailable in Turbo |
| Open Weights | H3-Base available | No open-weight release stated on the current product page |
| Speed | Usually 2–5 minutes | Usually 10–30 seconds; queue time may vary |
Choose the workflow that fits your resolution, control, and iteration needs.
Credit Pricing for MiniMax H3
Buy credits once and use them across supported MiniMax H3, H3 Max, and H3 Max Turbo workflows. No subscription, no auto-renewal, and purchased credits never expire.
5-second MiniMax H3 generation: 25 credits at 2K · 15 credits at 768pStarter
$9.90· 200 credits
8 five-second 2K videos $1.24 each
Creator
$29.90· 650 credits
26 five-second 2K videos $1.15 each
Studio
$99.90· 2,500 credits
100 five-second 2K videos $1.00 each
Failed or rejected generations are automatically refunded. Credit use varies by model, duration, resolution, and workflow.
MiniMax H3 FAQ
Learn how MiniMax H3 handles video length, resolution, audio, references, aspect ratios, pricing, and different generation workflows.
What resolution and video length does MiniMax H3 support?
This workspace supports 5- to 15-second videos at up to 2K resolution and 24 fps. Available settings may vary by generation mode.
What can I use as input for MiniMax H3?
Start with a text prompt, image, first and last frames, or multimodal references. Depending on the workflow, you can use images, video, and audio to guide appearance, motion, camera behavior, timing, and sound.
Does MiniMax H3 generate audio with the video?
Yes. H3 can generate synchronized stereo audio as part of the same workflow, including dialogue, music, ambience, and sound effects based on your prompt.
Which MiniMax H3 workflow should I use?
Use Text to Video when starting from an idea, Image to Video when you already have a visual starting point, First & Last Frame when you need a controlled beginning and ending, and Reference to Video when appearance, motion, timing, or audio needs more precise guidance.
How many reference files can I use?
The current workspace supports up to 9 image references, 3 video references, and 3 audio references. Include at least one image or video. Give each reference a specific role in the prompt rather than adding files that do not directly affect the shot.
Which aspect ratios does MiniMax H3 support?
Supported workflows include 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16. Use widescreen formats for cinematic and web video, square formats for feed content, and vertical formats for mobile-first video.
How much does a MiniMax H3 video cost?
A 5-second MiniMax H3 generation uses 15 credits at 768p or 25 credits at 2K without additional reference charges. Depending on the credit pack, a 5-second 2K video costs approximately $1.00–$1.24. Credit use varies by model, duration, resolution, and reference inputs. Review the live estimate before generating.See full credit pricing →
What happens if a generation fails?
Failed or rejected generations are automatically refunded, so credits are only kept for successfully delivered generations.
Can MiniMax H3 generate readable text and product details?
It can reproduce text, logos, packaging, and other visual details, but small lettering and complex labels may still change between frames. Keep important text short and review the final video carefully before publishing it.
How do I get better results from MiniMax H3?
Start with one clear subject and action, then describe the camera movement, lighting, visual style, and sound. Specify any details that must remain consistent, and change one instruction at a time when refining the result.Read the MiniMax H3 Prompt Guide →
Create Your First MiniMax H3 Video
Start with a prompt, image, or keyframe and generate your video directly in the browser.
Generate a VideoNo credit card required · Start with a prompt