VAE
ae.safetensors
Ae · VAE for ComfyUI
- File
ae.safetensors- Type
- VAE
- Size
- 335 MB
- Folder
ComfyUI/models/vae/- Source
- Comfy-Org/Z-Image-Turbo
- SHA-256
afc8e28272cd…
43 workflows use this file
What is Ae?
Ae is a VAE (Variational Autoencoder) that encodes and decodes latent representations. You can run it locally in ComfyUI with full control over every parameter, or access it through Comfy Cloud. ComfyUI's node-based workflow editor lets you connect Ae with other models and custom nodes to build any pipeline you need. There are 43 official ComfyUI workflow templates that use Ae, ready to load and customize.
Read the full tutorial →Workflows for this file
- Flux, Flux.1
Flux.1 Inpaint
Uses this file - Z-Image-Turbo, SDXL
Skin Enhancer + Upscaler
Uses this file - Flux, Flux.1
Flux.1 Dev: Text to Image
Uses this file - Flux.1 Kontext, Flux
Flux Kontext Dev Image Edit
Uses this file - Flux.1, Flux
Flux.1 Outpaint
Uses this file - Chroma, Flux
Chroma: Text to Image
Uses this file - Flux.1, Flux
Flux.1 Dev fp8: Text to Image
Uses this file - Flux, Flux.1 Krea Dev
Flux.1 Krea Dev
Uses this file
Frequently Asked Questions
Ae is a VAE (Variational Autoencoder) that encodes and decodes latent representations. You can run it locally in ComfyUI with full control over every parameter, or access it through Comfy Cloud. ComfyUI's node-based workflow editor lets you connect Ae with other models and custom nodes to build any pipeline you need. There are 43 official ComfyUI workflow templates that use Ae, ready to load and customize.
Follow the step-by-step tutorial at https://docs.comfy.org/tutorials/image/z-image/z-image-turbo. You can also load any of the 43 official ComfyUI workflow templates that use Ae directly in ComfyUI.
There are 43 official ComfyUI workflow templates that use Ae. Each template is ready to run in ComfyUI and can be customized to suit your project.
ComfyUI is free and open source. Ae weights are available to download from Hugging Face. You only pay for compute when running on Comfy Cloud; local inference on your own hardware is always free.