Skip to content

HunyuanVideo

An open framework for large video generation models

Visit official site

What HunyuanVideo is

The framework is described as an open-source Tencent system for training large video generation models, using data curation and joint image-video training.

What you can do with it

  • Use a unified image and video Transformer design
  • Rely on an MLLM text encoder and 3D VAE
  • Rewrite prompts and generate at several resolutions

Getting started

  • Clone the repository to a local machine
  • Set up a Conda env from environment.yml
  • Install pip packages and flash attention v2

Pros & cons

✓ What we like

  • Code and model weights are open
  • Multiple resolutions and frame rates are supported

! What to watch out for

  • Setup needs a Conda environment
  • flash attention v2 is installed for speed

FAQ

What is released?

The material states code and model weights are open sourced.

How large is the model?

The material states a video generation model with over 13 billion parameters.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review