---
author: cogvideox996
baseModel: CogVideoX
hashes:
  AutoV1: 8A96607E
  AutoV2: A410E48D98
  AutoV3: 37ABB2444528
  BLAKE3: F475F5AD96B6BB2A791A0DDD2C055810E4C12A4343CDD926BC2ED3B9ACE25FCD
  CRC32: 1AB3FB12
  SHA256: A410E48D988C8224CEF392B68DB0654485CFD41F345F4A3A81D3E6B765BB995E
metadata:
  format: SafeTensor
modelPage: https://civitai.com/models/1009676?modelVersionId=1131953
preview:
- https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/39402559-7cc5-4e33-b8ff-895ad0769b60/width=450/43812071.jpeg
website: Civitai
---

# Trigger Words

No trigger words

# About this version

* **SHA256:** a410e48d988c8224cef392b68db0654485cfd41f345f4a3a81d3e6b765bb995e
* **Pointer size:** 134 Bytes
* **Size of remote file:** 862 MB

# CogVideoX-VAE

The model is released under a permissive CogVideoX license.

To get started with ComfyUI;

1. Update to the latest version of ComfyUI
2. Download CogVideoX model weights into `models/diffusion_models` folder, use "CogVideoXModelLoader" load
3. Make sure a text encoder [[1](https://civitai.com/models/497255?modelVersionId=568405)] is in your `models/clip` folder
4. Download the VAE to: `ComfyUI/models/vae, use "`CogVideoXVAELoader`" load`

  
CogVideoX has [kijai ComfyUI support](https://github.com/kijai/ComfyUI-CogVideoXWrapper), and will run on 20GB+ or 9GB VRAM.

Github: <https://github.com/THUDM/CogVideo>

HuggingFace: <https://huggingface.co/collections/THUDM/cogvideo-66c08e62f1685a3ade464cce>