vizart-vj/ComfyUI-MiniMax-H3-LongMedia

ComfyUI custom nodes for MiniMax H3 long-form video/audio generation with streamed Sol attention, compressed KV, adaptive VRAM guards, chunked MLP/final output, and SAFE long-sequence optimizations for limited VRAM.

What it solves

This project provides a bridge to use MiniMax's H3-LongMedia model within the ComfyUI ecosystem, enabling users to generate long-form media content using specialized AI models in a node-based workflow.

How it works

It functions as a custom node extension for ComfyUI. By installing this package, users gain access to specific nodes that interface with the MiniMax H3-LongMedia API, allowing the model to be integrated into existing image or video generation pipelines.

Who it’s for

It is designed for ComfyUI users who want to leverage MiniMax's specific long-media generation capabilities within their local creative workflows.

Highlights

  • Enables integration of MiniMax H3-LongMedia models into ComfyUI.
  • Provides custom nodes for streamlined media generation.
  • Supports modular, node-based AI workflows.

Related

  • Project
    cncf/k8s-ai-conformance

    A certification program that defines standardized capabilities for Kubernetes platforms to ensure AI and machine learning workloads are portable and reliable across different clusters.

  • Project
    walkingddd/CPA-Helper

    A local self-hosted admin panel for CLIProxyAPI that provides usage analytics, user balance management, and API key lifecycle control.

  • Project
    ZJU-REAL/EasySteer

    A high-performance framework built on vLLM for steering LLM behavior using vectors in the hidden-state space without changing model weights.

  • Project
    aliemo/transfomers-silicon-research

    A curated, auto‑updated catalog of papers that implement Transformer‑style models on ASICs, FPGAs, and other silicon accelerators, with a searchable static website and a small admin UI for community curation.