Sign In
Documentation

ComfyUI

Deploy ComfyUI locally, then manually install the LLMPool MiniMax H3 video custom node package.

Install

Download the LLMPool ComfyUI MiniMax H3 plugin, stop ComfyUI, and extract it into ComfyUI/custom_nodes/. The final path must be ComfyUI/custom_nodes/ComfyUI-LLMPool-MiniMax/__init__.py.

Set LLMPOOL_API_KEY to a Board API key and LLMPOOL_MINIMAX_BASE_URL to the LLMPool HTTP(S) service root before starting ComfyUI. The plugin appends /minimax/v2 automatically; do not use the OpenAI /openai/v1 path. Do not put the API key in a node or workflow JSON.

Start ComfyUI with python main.py, open http://127.0.0.1:8188, and find the nodes in video/LLMPool:

  • LLMPool MiniMax H3 文生视频 generates a video from text.
  • LLMPool MiniMax H3 首帧生成视频 uses one ComfyUI image as the first frame.

The nodes poll the task, download the completed MP4 into ComfyUI's output directory, and expose preview/download metadata.