All models
Video generationLive
falMiniMax H3 Max
fal's speed-focused post-trained version of MiniMax H3 for short video with native audio, fast 768p iteration, and text, image, or multimodal reference input.
Plain-English overview
What MiniMax H3 Max actually is
H3 Max starts from MiniMax's open-weight H3 foundation, then adds fal's post-training data and an inference stack designed around prompt adherence, visual appeal, and throughput. The result is a hosted fal model rather than a new checkpoint released by MiniMax itself.
The model can create five- to fifteen-second clips from a prompt, animate a first or end frame, or use image, video, and audio references. Its native working resolution is 768p; the 1080p option uses latent refinement, while MiniMax's separate full H3 system is the route associated with a 2K regeneration stage.
Good fit for
- Rapid video ideation with native sound
- Image-to-video where fast iteration matters
- Reference-guided short clips and creative testing
Category comparison
The facts that matter for video models
These are provider-published specifications, not Cody benchmark scores. Follow the linked sources for current limits and endpoint-specific exceptions.
- Generation price
- $0.05 480p · $0.08 768p · $0.16 1080p / secCurrent list price per generated second or the provider's closest billing unit.
- Clip duration
- 5–15 secondsDocumented duration choices or maximum clip length.
- Maximum output
- 1080p latent refinement; 768p nativeHighest provider-documented output resolution, with relevant constraints.
- Native audio
- YesWhether the endpoint generates synchronized audio with video.
- Input modes
- Text, first/end image, or image/video/audio referencesSupported text, image, video, reference, or motion inputs.
- Generation time
- fal test: ~2.46s inference for a 5s 768p clipOnly shown when a provider publishes a range; not a Cody benchmark.
Pricing & comparisons
Estimate your cost
Set your usage. Your estimate updates as you type.
Example: 8-second, 768p clips with audio. Models that do not support these settings will not receive an estimate.
MiniMax H3 Max
fal
Estimated total (USD)
For the usage above · USD · API pricing, not a subscription
How this estimate works
Estimates exclude taxes, tools, cache storage/writes, free allowances and custom discounts. Image estimates cover output only, not prompt or reference-image charges. Quality modes differ by model. Unlisted settings are not treated as free.
API and provider access
Where to get MiniMax H3 Max
falH3 Max text-to-video APIDirect prompt-to-video route with duration, resolution, aspect ratio, and prompt-expansion controls.falH3 Max image-to-video APIAnimate a first frame, an end frame, or both as keyframes.falH3 Max reference-to-video APIGuide a clip with image, video, or audio references; reference inputs can add token charges.
Availability
Regions and access stage
Available through fal's web experience and queued API routes for text-to-video, image-to-video, and reference-to-video.
fal does not publish one endpoint-specific processing-country list on the H3 Max pages. Check the live service terms and account requirements for your region.
Check live availabilityData and training
The route matters.
fal stores request inputs and outputs by default. The X-Fal-Store-IO: 0 header prevents payload storage, but separately uploaded CDN inputs can remain accessible and need their own deletion or lifecycle handling.
This is a concise reading of the cited provider material, not legal advice. A third-party gateway can have different storage, routing, training, and residency terms from the model maker's direct API.
Read the provider policyFrequently asked questions
MiniMax H3 Max FAQ
What is MiniMax H3 Max?
fal's speed-focused post-trained version of MiniMax H3 for short video with native audio, fast 768p iteration, and text, image, or multimodal reference input. H3 Max starts from MiniMax's open-weight H3 foundation, then adds fal's post-training data and an inference stack designed around prompt adherence, visual appeal, and throughput. The result is a hosted fal model rather than a new checkpoint released by MiniMax itself.
When was MiniMax H3 Max released?
MiniMax H3 Max was released on August 27, 2026 according to the cited provider materials.
Where can I access MiniMax H3 Max?
Available through fal's web experience and queued API routes for text-to-video, image-to-video, and reference-to-video. The access routes listed in this guide are fal.
How much does MiniMax H3 Max cost?
$0.05 480p · $0.08 768p · $0.16 1080p / second. These are fal's standard H3 Max rates effective September 15, 2026. A launch promotion discounted them by 75% through September 14. Reference-to-video can also bill reference tokens beyond the included allowance.
Where is MiniMax H3 Max available?
Available through fal's web experience and queued API routes for text-to-video, image-to-video, and reference-to-video. fal does not publish one endpoint-specific processing-country list on the H3 Max pages. Check the live service terms and account requirements for your region.
Is my MiniMax H3 Max API data used for training?
fal stores request inputs and outputs by default. The X-Fal-Store-IO: 0 header prevents payload storage, but separately uploaded CDN inputs can remain accessible and need their own deletion or lifecycle handling. The policy belongs to the provider route and account terms, so verify it again before production use.
Related comparisons
Head-to-head comparisons
MiniMax H3 Max vs Sora 2
OpenAI's legacy synced-audio video model for short text- or image-guided clips—still documented, but approaching a permanent API shutdown.
Open full comparisonMiniMax H3 Max vs Veo 3.1
Google's preview video model for text-, image-, and video-guided generation with native audio and output options up to 4K.
Open full comparisonMiniMax H3 Max vs Veo 3.1 Fast
Google's lower-cost Veo 3.1 variant for quickly creating short video clips with native audio and output options up to 4K.
Open full comparisonMiniMax H3 Max vs Runway Gen-4.5
Runway's flagship developer video model for text- and image-guided clips, with professional formats and a predictable credit-per-second base rate.
Open full comparisonMiniMax H3 Max vs Kling Video 3.0 Pro
Kuaishou's Pro video model delivered through fal for multi-shot text/image generation, optional native audio, references, and clips up to 15 seconds.
Open full comparisonMiniMax H3 Max vs Kling Video 2.5 Turbo Pro
A cost-conscious Kling model on fal for fluid five- or ten-second clips generated from text or a starting image.
Open full comparisonMiniMax H3 Max vs Grok Imagine Video 1.5
SpaceXAI's current video model for turning text, images, or references into short clips with optional generated speech and sound.
Open full comparisonMiniMax H3 Max vs Seedance 2.5
ByteDance's current audiovisual model for coherent clips up to 30 seconds, with native sound, long-form storytelling, rich references, and targeted editing.
Open full comparisonMiniMax H3 Max vs HappyHorse 1.1
Alibaba's unified video family for creating short clips with sound from text, a starting image, or a reference video.
Open full comparisonMiniMax H3 Max vs LTX-2.5 Pro
Lightricks' quality-focused video model for making 720p or 1080p clips from text, images, or audio with native sound.
Open full comparisonMiniMax H3 Max vs LTX-2.5 Fast
LTX's speed-optimized video model for text, image, or audio input, native sound, automatic duration, and output up to 4K.
Open full comparisonMiniMax H3 Max vs LTX-2.3 Pro
LTX's mature full-workflow video model for generation plus Retake, Extend, Reframe, native audio, and first-to-last-frame control.
Open full comparison
