H3cApp

English | 日本語

Create AI videos with sound, directly on your Mac. H3cApp uses MiniMax H3 to turn text, images and other media into short videos. After downloading a model, generation takes place on your Mac—no cloud video-generation service, Python, or complicated setup required.

Powered by MiniMax H3 — an independent app, not an official MiniMax product.

What you can do

• Text to video: Describe a scene and generate video with synchronized sound.
• Image to video: Choose a starting image, an ending image, or both.
• Reference-based video: Guide the result using images, videos or audio (optional Ref2VA model required).
• Lip sync: Create video matching a voice recording or song.
• Longer videos: Continue previous clips and join them automatically.
• Other features: Progress previews, project management, and compatible LoRAs.

Mac requirements

Item	Requirement
Mac	MacBook Air, MacBook Pro, Mac mini, iMac, Mac Studio, or Mac Pro with Apple Silicon
Chip	Apple M-series (M1 or later); M5 recommended for faster processing
Memory	24 GB or more recommended
macOS	macOS 15 (Sequoia) or later
Other software	None for the prebuilt app

Memory compatibility: Tested on an M5 Mac with 24 GB of unified memory. 16 GB has not been tested and is not guaranteed to work. Macs with less than 16 GB have not been tested either. Intel Macs are not supported.

M1–M4 Macs can also use the resident or SSD-streaming compute modes, but processing speed varies by hardware and settings.

Models and disk space

You download a MiniMax H3 model separately from the app.

Model configuration	Approximate model storage	Best for
GGUF standard (recommended)	45 GB	Balanced quality and storage
GGUF compact	35 GB	Lower disk usage
Official BF16 (FL2VA)	145 GB	Original unquantized model
Official BF16 (FL2VA + Ref2VA)	290 GB	Original model with reference-based generation

These figures are model storage sizes, not RAM requirements. The GGUF configurations support text-to-video and first/last-image generation. Reference images, videos or audio require an additional Ref2VA model (about 11–21 GB).

On supported M5 Macs, an optional speed-up cache uses about 19 GB of extra disk space per model; using Ref2VA as well adds another cache of about 19 GB for it. Leave additional space for downloads, temporary files and generated videos. Models can be stored on an external SSD.

See GGUF model details for the available model combinations.

Install and create your first video

1. Download the latest H3cApp .dmg from Releases.
2. Open the .dmg and drag H3cApp into Applications.
3. Open H3cApp. The setup wizard will help you download a model. GGUF standard is a good starting point.
4. Type a prompt, for example, “A quiet beach at sunset, gentle waves and warm light.” Choose the length and size, then click Generate.
5. Preview the result and use Export Video… to save it.

Generation can take several minutes or longer depending on the Mac and settings. No Xcode or command-line setup is needed for the downloadable app.

For screenshots and step-by-step instructions, visit the Getting Started guide.

Guides and developer documentation

• Continuing and joining videos
• Model downloads and GGUF imports
• User guide and advanced settings
• Technical reference · Local API · Building from source
• FAQ · Report an issue

Publishing videos and licenses

When publishing a generated video, clearly disclose that it was AI-generated, as required by MiniMax H3's usage policy. Export Video… can add “AI-generated with MiniMax H3” to the video (on by default). You can also use the platform's AI label or mention it in the description. The app includes AI-generation information in MP4 metadata.

H3cApp is under the MIT License, based on antirez/h3.c. The model has separate license terms; review the MiniMax H3 Community License before use. See third-party notices and technical metadata/license notes.