Instructions to use MiniMaxAI/MiniMax-H3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use MiniMaxAI/MiniMax-H3 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("MiniMaxAI/MiniMax-H3", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
MiniMax Director β a ComfyUI timeline editor that covers every documented H3 feature
I built a ComfyUI workflow that gives H3 a timeline instead of one prompt box, and I tried to cover every documented feature of the model.
Tracks for picture, camera and audio. Take a person, a prop, a place, a style or a voice out of any file β each becomes a card you paste into any segment, and it stays the same across shots. It compiles the six-section reference format, task types, dialogue in the model's exact <d> form, frame anchors and on-screen text, and it enforces H3's own rules before you run: 24 fps, clip length 5 + 17n frames, 9 images / 3 videos / 3 audio / 12 files. A report node lints the graph; a prompt node shows the exact text the model receives.
MIT, in ComfyUI Manager as minimax-director:
https://github.com/imbutus/ComfyUI-MiniMaxDirector
Full walkthrough, 34 min: https://youtu.be/XIKDfvoFcCY
