A fast AI Video Generator for the GPU Poor. Supports Wan 2.1/2.2, Qwen Image, Hunyuan Video, LTX Video and Flux.

Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model

Official implementation for "RIFLEx: A Free Lunch for Length Extrapolation in Video Diffusion Transformers" (ICML 2025) and "UltraViCo: Breaking Extrapolation Limits in Video Diffusion Transformers"

[ICML2025, NeurIPS2025 Spotlight] Sparse VideoGen 1 & 2: Accelerating Video Diffusion Transformers with Sparse Attention

[NeurIPS 2025] Radial Attention: O(nlogn) Sparse Attention with Energy Decay for Long Video Generation

https://wavespeed.ai/ Context parallel attention that accelerates DiT model inference with dynamic caching

Less is Enough: Training-Free Video Diffusion Acceleration via Runtime-Adaptive Caching

Text Encoders finally matter 🤖🎥 - scale CLIP & LLM influence! + a Nerdy Transformer Shuffle node

Gradio UI for training video models using finetrainers

ComfyUI-HunyuanVideo-Avatar is now available in ComfyUI, HunyuanVideo-Avatar is a multimodal diffusion transformer (MM-DiT)-based model capable of simultaneously generating dynamic, emotion-controllable, and multi-character dialogue videos.

Custom tool set mostly for Hunyuan Video, but includes some WAN and FramePack Video nodes.

ComfyUI-HunyuanPortrait is now available in ComfyUI, HunyuanPortrait is a diffusion-based condition control method that employs implicit representations for highly controllable and lifelike portrait animation.

Avernus API server and client for LLM, SDXL, FLUX, Qwen-Image, WAN, and Ace-Step inference and many other architectures.

Benchmarking training throughput of contemporary diffusion models on AMD's GPU hardware.

HunyuanVideo API, 8bit version

AI Text-to-Video Generation Example using Hunyuan Model

Hunyuan-video inference with ComfyUI

HunyuanVideo 1.5 Generator