⚠️Note: NSFW content is not supported.
Optional — defaults to 16:9
⚠️Note: NSFW content is not supported.
Reference media (optional)
Your MiniMax H3 Video Awaits
Type a prompt, pick duration and resolution, optionally add reference media, then hit Generate.
Mix Five Inputs. Render One Cinematic Clip.
MiniMax H3 is our most flexible video AI. Drop in a text prompt, a first-frame image, a last-frame image, up to 10 reference videos, and audio — combine them any way you like and MiniMax H3 builds a 4–15 second short at 768P or 2K. One model, every input type, no trade-offs.
How MiniMax H3 Builds a Video From Mixed Inputs
Combine any inputs, pick a duration, and let MiniMax H3 do the rest. No single-input-only pipeline — every combination is supported.
Pick Your Input Mix
Start with text alone, drop a first-frame image, anchor a last-frame image for transition control, or combine multiple input types. MiniMax H3 understands your creative intent across every format.
Add Reference Media
Layer reference images for style, reference videos for motion, and audio for rhythm. MiniMax H3 reads every reference simultaneously, no manual switching between modes.
Render at 768P or 2K
Pick duration (4–15 seconds) and resolution (768P for speed, 2K for broadcast). Click Generate. MiniMax H3 produces the final MP4 in seconds.
What Makes MiniMax H3 Different
A single video AI engine that accepts every input type simultaneously — without forcing you to choose one mode over another.
Text to Video Built In
Describe any scene and MiniMax H3 brings it to life. 4–15 second cinematic clips from a single prompt, no setup required.
First-Frame Animation
Provide a starting image and MiniMax H3 animates it naturally. Useful for storyboarding and extending existing concepts into video.
Last-Frame Transition Control
Set your ending image and MiniMax H3 builds a smooth transition from your first frame to your last — useful for morphs and reveals.
Reference Video Motion
Drop reference videos and MiniMax H3 follows the motion, pacing, and camera style across every output.
Audio Synchronized Visuals
Upload audio and MiniMax H3 syncs the rendered motion to your music or voice — ideal for music videos and narration pieces.
Multiple Reference Images
Blend up to 10 reference images for style transfer and creative compositions. MiniMax H3 combines every reference into a coherent output.
Platform-Optimized Aspect Ratios
16:9 for YouTube, 9:16 for TikTok and Reels, plus more. MiniMax H3 picks the right composition for the destination platform.
Up to 2K Output
Choose 768P for fast drafts or 2K for high-end placements. MiniMax H3 hits broadcast-grade quality at the top tier.
MiniMax H3 Pricing — Pay Per Render
Only pay for what you generate. 70% gross margin on every MiniMax H3 render, no subscription required.
| Duration | 768P — Standard | 2K — Pro |
|---|---|---|
| 4s | 115 credits | 186 credits |
| 5s | 143 credits | 233 credits |
| 7s | 200 credits | 325 credits |
| 10s | 286 credits | 464 credits |
| 15s | 429 credits | 697 credits |
From 115 credits per 4-second clip at 768P
- Text-only, first-frame, last-frame, reference image / video / audio
- Free credits for new sign-ups
- Full commercial rights on paid plans
- Refunded automatically on generation failure
What creators say about MiniMax H3
Real feedback from creators and agencies using MiniMax H3 daily.
“MiniMax H3 finally combines text and images naturally in one video AI. Shipped 30 campaign clips in one afternoon.”
“Reference video guidance is the killer feature — MiniMax H3 actually follows the motion style I drop in.”
“First-frame + last-frame control let me storyboard entire scenes before MiniMax H3 renders them.”
“2K MiniMax H3 output looks premium for product videos. Clients have stopped asking for reshoots.”
“Started with free credits on MiniMax H3, now I'm on the Pro plan. The multimodal mix is worth it.”
“Audio sync on MiniMax H3 is wild — uploaded a track, got a full music video back.”
“Switched off a text-only generator for MiniMax H3. Reference video guidance changed my workflow.”
“Agency runs 200+ MiniMax H3 renders per day. The mixed-input pipeline never chokes.”
“2K MiniMax H3 output saved a campaign launch. Clients were blown away by the fidelity.”
“My students use MiniMax H3 for explainers — visual storytelling clicks faster with mixed inputs.”
MiniMax H3 — Common Questions
Quick answers about MiniMax H3's multimodal video pipeline.
What is MiniMax H3?
MiniMax H3 is our flagship multimodal video AI. It accepts text, images, videos, and audio simultaneously — combining every input type in a single generation for finer creative control than single-input tools allow.
Which input modes does MiniMax H3 support?
Text-only prompts, first-frame image animation, last-frame image control, reference video guidance, reference image styling, and audio synchronization. Use any combination in one MiniMax H3 generation.
Is MiniMax H3 free to try?
New accounts receive starter credits — enough for several MiniMax H3 generations. Pay-as-you-go pricing kicks in after that.
How fast is MiniMax H3?
A 5-second MiniMax H3 render finishes in 30–60 seconds. The longest 15-second outputs land in under two minutes.
Can I use both first-frame and last-frame images in MiniMax H3?
Yes. Setting both gives you precise control over the start and end of the video — MiniMax H3 builds a smooth transition between them.
What resolutions are available?
768P for faster drafts or 2K for the highest quality. Both resolutions are available for every MiniMax H3 generation.
How long can each MiniMax H3 video be?
4 to 15 seconds per render. For longer narratives, render multiple clips and stitch them in your editor.
Can I add audio to my MiniMax H3 render?
Yes. Upload audio (MP3, WAV, FLAC) and MiniMax H3 will sync the visuals to your track's rhythm and energy.
Can I use MiniMax H3 outputs commercially?
Yes. Paid plans include full commercial rights on every MiniMax H3 render — use them in ads, courses, client work, and paid social.
How is MiniMax H3 different from text-only video tools?
MiniMax H3 combines multiple input types in a single video AI. While text-only video generation works fine, layering reference images, videos, and audio gives you much finer control over the final result.
Single-input video generators force trade-offs. MiniMax H3 combines text, images, reference videos, and audio in any combination — one model, every input type.