MiniMax H3 video and audio generation workflows in ComfyUI inside Promptus
Claudia Perez
AI Video

MiniMax H3 in ComfyUI: Install It Locally

Promptus
October 28, 2026
wiki 305
promptus ai video generator

MiniMax H3 in ComfyUI: Install It Locally with 3 Ready-Made Workflows

Short answer: MiniMax H3 makes video and its audio together — voices, music and sound effects in one pass, for clips up to 15 seconds. In Promptus it comes as three ready-made CosyTemplates — text to video, image to video and reference to video — that you install with Install Locally. Check two things first: the files are large, and its licence excludes the EU, the UK, South Korea and the US.

Licence check before you use MiniMax H3: its Community License grants rights only outside the European Union, the United Kingdom, South Korea and the United States, which it lists as excluded territories; people there have to contact MiniMax for a licence. Commercial products earning more than $20 million a year need MiniMax’s written authorisation, and commercial products must show “MiniMax H3” in their interface. Read the licence for your own case.

The three workflows

WorkflowWhat you give itUse it for
Text to VideoA promptDescribe the scene, the camera movement and the sound you want
Image to VideoYour own picture plus a promptStart from your image and describe the motion
Reference to VideoUp to 9 images, 3 videos and 3 audio clips (12 files max)Guide the character, the motion and the voice

How to install it in Promptus

  1. Open CosyTemplates and search for MiniMax H3.
  2. Choose the workflow you need, open the card’s menu and click Install Locally. Promptus downloads every part of the kit into the right folders.
  3. When it moves to Available Locally, choose Load in ComfyUI.
  4. Write the prompt: for text to video, the scene, camera and sound; for image to video, the motion only, because the picture is already the first frame.

If Install Locally is greyed out, start the local ComfyUI server from Server in the sidebar first. In Promptus’ own testing, MiniMax H3 failed in Playground with a SaveVideo ‘format’ error and worked when run from the ComfyUI canvas — so use Load in ComfyUI.

File sizes

MiniMax H3 is big. The Comfy-Org repack lists these files:

Partbf16Prunedint8
Video model (each of first-last-frame and reference)66.3 GB40.2 GB34.0 GB
Text encoder (Qwen3-VL 32B)51.5 GB—27.1 GB

Which files a template uses depends on the template, so check free disk space before installing.

How fast is it?

In a separate test of five local image-to-video models on an RTX 5090 Laptop GPU with 24 GB, MiniMax H3 took about 3.1 minutes per 5-second clip at 640×640 and 24 fps — and was the only one of the five to produce an audio track. In the AI influencer walkthrough, it gave a character her voice, with each 5-second clip taking about 4 to 6 minutes.

Writing prompts for sound

  • Write the scene like a director: what happens, what the camera does, and what we hear.
  • Put spoken words in quotation marks.
  • Describe the background sound too — music, ambience, effects.

The licence, in plain terms

  • Rights are granted only outside the excluded territories: the European Union, the United Kingdom, South Korea and the United States.
  • Commercial products and services with more than $20 million in yearly revenue need MiniMax’s prior written authorisation.
  • Commercial products using it must prominently show “MiniMax H3” in their interface.
  • Redistributing the model requires passing on the licence and a NOTICE file.

This is a summary, not legal advice — read the licence for your own case.

Links

Download Promptus for Mac and Windows.

Frequently Asked Questions

What is MiniMax H3?

A video model from MiniMax that makes the video and its audio together, including voices, music and sound effects, in one pass, for clips up to 15 seconds.

Which MiniMax H3 workflows are in Promptus?

Three ready-made CosyTemplates: Text to Video, Image to Video and Reference to Video.

How many reference files can MiniMax H3 use?

Reference to Video takes up to 9 images, 3 videos and 3 audio clips, with 12 files at most.

How big are the MiniMax H3 model files?

In the Comfy-Org repack, each full bf16 video model file is 66.3 GB, with pruned versions at 40.2 GB and int8 versions at 34.0 GB, and the Qwen3-VL 32B text encoder is 51.5 GB in bf16 or 27.1 GB in int8.

Can I use MiniMax H3 in the US, UK or EU?

Its Community License grants rights only outside the European Union, the United Kingdom, South Korea and the United States. People there have to contact MiniMax for a licence.

Why does MiniMax H3 fail in Playground?

In Promptus testing it failed in Playground with a SaveVideo format error. Running the same template from the ComfyUI canvas with Load in ComfyUI worked.

Written by:
Claudia Perez
Claudia Perez is a tech creator and founder at Promptus. She loves to design and share expressive image workflows. Her work focuses on visual storytelling, refined aesthetics, and ComfyUI workflows.
Try Promptus Cosy UI today for free.
ai image generator

AI Generation Platform

Promptus AI is the easiest way to generate realistic photos, videos, 3D and ComfyUI workflows with artificial intelligence.

Our AI photo generator produces lifelike portraits, product images, and creative concepts in seconds, making it the perfect tool for creators and brands.

promptus ai video generator