Skip to content

CogVideoX-2B

An open source video generation model

Visit official site

What CogVideoX-2B is

The model is described as an open source video generator developed by a Tsinghua University team. It generates video from English prompts and is deployed with the Hugging Face diffusers library.

What you can do with it

  • Generate video from an English prompt
  • Produce a 6-second clip
  • Use sinusoidal position embeddings

Model notes

  • 36GB of GPU memory is needed
  • Output is 720 by 480 at 8 fps
  • Quantised inference is unsupported

Pros & cons

✓ What we like

  • Deployment uses the Hugging Face diffusers library
  • The model is open source

! What to watch out for

  • 36GB of GPU memory is needed for inference
  • Quantised inference and multi-GPU inference are not supported

FAQ

How long is the output?

The card states 6 seconds at 8 frames per second.

What resolution is used?

720 by 480.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review