Install ComfyUI on Mac OS for Flux 1

This post serves as a document recording how to install ComfyUI on Mac OS. This is intended to act as a guide for anyone who wishes to install ComfyUI on their Apple Mac and needs a bit of a beginner-ish friendly step-by-step process. This process will setup you up for running Flux 1 on your Apple Mac and diverges from other guides as you will need to have the Nightly version of PyTorch installed to make this work.

Laying the Foundations

I’ll assume that you are starting afresh and don’t have the neccessary prerequisites. Feel free to move on if you already have these in place, but it seems like a good place to start.

First off install Homebrew

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

First off, make sure you have Git installed.

brew install git

Next up install Python3

brew install python@3.13

To make your version of ComfyUI Flux 1 ready you first off need to install the nightly build of PyTorch. You do this by running the following command in the terminal:

pip3 install --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/cpu

Older versions of PyTorch may prevent Flux from working, so this is defintiely the way to go.

Clone the Git ComfyUI Repository

git clone https://github.com/comfyanonymous/ComfyUI

This will create a ComfyUI folder in your users home directory. You can see this with the following command:

ls

Within the terminal go into the change directory to the ComfyUI folder which has been created.

cd ComfyUI

It’s important to note the capital letters…

Now install using the following command

pip3 install -r requirements.txt

Once complete you can re-start ComfyUI

python3 main.py

Then you can open up ComfyUI by going to this location in your browser

http://127.0.0.1:8188

There are no models installed, but rather than go and download, it’s best to install ComfyUI Manager as this helps with the admin of the tool, so exit the browser and press the following to terminate ComfyUI

Ctrl+c

change directory to go into custom_nodes

cd custom_nodes

And then git clone the ComfyUI Manager repo:

git clone https://github.com/ltdrdata/ComfyUI-Manager comfyui-manager

restart ComfyUI by going back to ComfyUI

cd ..

And then restart the application

python3 main.py

Now when you open the application you will have the Manager tool installed. This will allow you to install the v1-5-pruned-emaonly.ckpt model. Which is a base Stable Diffussion model.

ComfyUI Manager button appears in the top right hand corner of the browser:

Go into the Model Manager and install the v1-5-pruned-emaonly.ckpt. This is a 4.27GB file and so will take some time to download.

Once the 4.27GB file has downloaded you should restart as prompted by ComfyUI Manager and then you are ready to do your first image generation. To do this click on the Queue button in the bottom right hand corner…

If you see this error about not having the model installed

You should refresh your browser window and then click on the ckpt_name in the Load Checkpoint node to select the model you’ve just downloaded.

Here’s what your first image generated locally on your Mac is likely to look like:

Don’t worry too much about the quality of the image as in the next in the series will walk you through setting up Flux 1 and building a workflow which will give you much higher quality images.