How to install ComfyUI manually in different systems

For the installation of ComfyUI, it is mainly divided into several steps:

1. Create a virtual environment(avoid polluting the system-level Python environment)
2. Clone the ComfyUI code repository
3. Install dependencies
4. Start ComfyUI

Python 3.13 is very well supported. If you have trouble with some custom node dependencies on 3.13 you can try 3.12.

Python 3.14 will work if you comment out the kornia dependency in the requirements.txt file (breaks the canny node) but it is not recommended.

You can also refer to ComfyUI CLI to install ComfyUI, it is a command line tool that can easily install ComfyUI and manage its dependencies.

Create a virtual environment

Independent virtual environments are necessary because ComfyUI’s dependencies may conflict with other dependencies on the system, and it can also avoid polluting the system-level Python environment.

Install Miniconda. This will help you install the correct versions of Python and other libraries needed by ComfyUI.

Create an environment with Conda.

conda create -n comfyenv
conda activate comfyenv

Clone the ComfyUI code repository

You need to ensure that you have installed Git on your system. First, you need to open the terminal (command line), then clone the code repository.

Windows

If you have not installed Microsoft Visual C++ Redistributable, please install it here.

git clone [e-mail]:comfyanonymous/ComfyUI.git

Install GPU and ComfyUI dependencies

1. Install GPU dependencies

   Install GPU Dependencies

   Nvidia

   Nvidia users should install stable pytorch using this command:

   pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu130
   

   Alternatively, you can install the nightly version of PyTorch.

   Install Nightly

   Install Nightly version (might be more risky)

   This is the command to install pytorch nightly instead which might have performance improvements:

   pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu130
   

   Troubleshooting

   If you get the “Torch not compiled with CUDA enabled” error, uninstall torch with:

   pip uninstall torch
   

   And install it again with the command above.

   AMD (Linux)

   AMD users can install rocm and pytorch with pip if you don’t have it already installed, this is the command to install the stable version:

   pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm6.4
   

   Alternatively, you can install the nightly version of PyTorch.

   Install Nightly

   Install Nightly version (might be more risky)

   This is the command to install the nightly with ROCm 7.0 which might have some performance improvements:

   pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm7.0
   

   AMD Experimental (Windows and Linux)

   Experimental support for Windows and Linux. RDNA 3, 3.5 and 4 only.

   These have less hardware support than the builds above but they work on Windows. You also need to install the pytorch version specific to your hardware.

   RDNA 3 (RX 7000 series):

   pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx110X-dgpu/
   

   RDNA 3.5 (Strix halo/Ryzen AI Max+ 365):

   pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx1151/
   

   RDNA 4 (RX 9000 series):

   pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx120X-all/
   

   Intel

   Intel Arc GPU users can install native PyTorch with torch.xpu support using pip. More information can be found here.

   Option 1: Install PyTorch xpu

   To install PyTorch xpu, use the following command:

   pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/xpu
   

   Install Nightly

   Install Nightly version (might be more risky)

   This is the command to install the Pytorch xpu nightly which might have some performance improvements:

   pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/xpu
   

   Option 2: Intel Extension for PyTorch (IPEX)

   Alternatively, Intel GPUs supported by Intel Extension for PyTorch (IPEX) can leverage IPEX for improved performance.

   Visit Installation for more information.

   Mac ARM Silicon

   You can install ComfyUI in Apple Mac silicon (M1 or M2) with any recent macOS version.

   Install pytorch nightly. For instructions, read the Accelerated PyTorch training on Mac Apple Developer guide (make sure to install the latest pytorch nightly).

   conda install pytorch-nightly::pytorch torchvision torchaudio -c pytorch-nightly
   

   Remember to add your models, VAE, LoRAs etc. to the corresponding Comfy folders (models/checkpoints, models/vae, etc.).

   Ascend NPUs

   For models compatible with Ascend Extension for PyTorch (torch_npu). To get started, ensure your environment meets the prerequisites outlined on the installation page.

   Step-by-step guide:

   1. Begin by installing the recommended or newer kernel version for Linux as specified in the Installation page of torch-npu, if necessary.
   2. Proceed with the installation of Ascend Basekit, which includes the driver, firmware, and CANN, following the instructions provided for your specific platform.
   3. Next, install the necessary packages for torch-npu by adhering to the platform-specific instructions on the Installation page.
   4. Finally, follow the ComfyUI manual installation guide for Linux. Once all components are installed, you can run ComfyUI as described earlier.

   Cambricon MLUs

   For models compatible with Cambricon Extension for PyTorch (torch_mlu).

   Step-by-step guide:

   1. Install the Cambricon CNToolkit by adhering to the platform-specific instructions on the Installation
   2. Next, install the PyTorch(torch_mlu) following the instructions on the Installation
   3. Launch ComfyUI by running python main.py

   Iluvatar Corex

   For models compatible with Iluvatar Extension for PyTorch.

   Step-by-step guide:

   1. Install the Iluvatar Corex Toolkit by adhering to the platform-specific instructions on the Installation
   2. Launch ComfyUI by running python main.py
2. Install ComfyUI dependencies

   cd ComfyUI
   pip install -r requirements.txt
   

3. Start ComfyUI

   Start the application

   cd ComfyUI
   python main.py
   

How to update ComfyUI

1. pull the latest code

   Use the command line to enter the installation path of ComfyUI, then pull the latest code.

   cd <installation path>/ComfyUI
   git pull
   

2. install the dependencies

   Use the command line to enter the installation path of ComfyUI, then install the dependencies.

   You need to ensure that the current Python environment is the ComfyUI virtual environment, otherwise the dependencies will be installed to the system-level Python environment, polluting the system-level Python environment.

       pip install -r requirements.txt
   

Adding Extra Model Paths

If you want to manage your model files outside of ComfyUI/models, you may have the following reasons:

• You have multiple ComfyUI instances and want them to share model files to save disk space
• You have different types of GUI programs (such as WebUI) and want them to use the same model files
• Model files cannot be recognized or found

We provide a way to add extra model search paths via the extra_model_paths.yaml configuration file

Open Config File

Portable/Manual Install

For the ComfyUI version such as portable and manual, you can find an example file named extra_model_paths.yaml.example in the root directory of ComfyUI:

ComfyUI/extra_model_paths.yaml.example

Copy and rename it to extra_model_paths.yaml for use. Keep it in ComfyUI’s root directory at ComfyUI/extra_model_paths.yaml. You can also find the config example file here

If the file does not exist, you can create it yourself with any text editor.

Example Structure

Suppose you want to add the following model paths to ComfyUI:

📁 YOUR_PATH/
  ├── 📁models/
  |   ├── 📁 lora/
  |   │   └── xxxxx.safetensors
  |   ├── 📁 checkpoints/
  |   │   └── xxxxx.safetensors
  |   ├── 📁 vae/
  |   │   └── xxxxx.safetensors
  |   └── 📁 controlnet/
  |       └── xxxxx.safetensors

Then you can configure the extra_model_paths.yaml file like below to let ComfyUI recognize the model paths on your device:

my_custom_config:
    base_path: YOUR_PATH
    loras: models/loras/
    checkpoints: models/checkpoints/
    vae: models/vae/
    controlnet: models/controlnet/

or

my_custom_config:
    base_path: YOUR_PATH/models/
    loras: loras
    checkpoints: checkpoints
    vae: vae
    controlnet: controlnet

For the desktop version, please add the configuration to the existing configuration path without overwriting the path configuration generated during installation. Please back up the corresponding file before modification, so that you can restore it when you make a mistake.

Or you can refer to the default extra_model_paths.yaml.example for more configuration options. After saving, you need to restart ComfyUI for the changes to take effect.

Below is the original config example:

#Rename this to extra_model_paths.yaml and ComfyUI will load it


#config for a1111 ui
#all you have to do is change the base_path to where yours is installed
a111:
    base_path: path/to/stable-diffusion-webui/

    checkpoints: models/Stable-diffusion
    configs: models/Stable-diffusion
    vae: models/VAE
    loras: |
         models/Lora
         models/LyCORIS
    upscale_models: |
                  models/ESRGAN
                  models/RealESRGAN
                  models/SwinIR
    embeddings: embeddings
    hypernetworks: models/hypernetworks
    controlnet: models/ControlNet

#config for comfyui
#your base path should be either an existing comfy install or a central folder where you store all of your models, loras, etc.

#comfyui:
#     base_path: path/to/comfyui/
#     # You can use is_default to mark that these folders should be listed first, and used as the default dirs for eg downloads
#     #is_default: true
#     checkpoints: models/checkpoints/
#     clip: models/clip/
#     clip_vision: models/clip_vision/
#     configs: models/configs/
#     controlnet: models/controlnet/
#     diffusion_models: |
#                  models/diffusion_models
#                  models/unet
#     embeddings: models/embeddings/
#     loras: models/loras/
#     upscale_models: models/upscale_models/
#     vae: models/vae/

#other_ui:
#    base_path: path/to/ui
#    checkpoints: models/checkpoints
#    gligen: models/gligen
#    custom_nodes: path/custom_nodes

For example, if your WebUI is located at D:\stable-diffusion-webui\, you can modify the corresponding configuration to

a111:
    base_path: D:\stable-diffusion-webui\
    checkpoints: models/Stable-diffusion
    configs: models/Stable-diffusion
    vae: models/VAE
    loras: |
         models/Lora
         models/LyCORIS
    upscale_models: |
                  models/ESRGAN
                  models/RealESRGAN
                  models/SwinIR
    embeddings: embeddings
    hypernetworks: models/hypernetworks
    controlnet: models/ControlNet

Add Extra Custom Nodes Path

Besides adding external models, you can also add custom nodes paths that are not in the default path of ComfyUI

Please note that this will not change the default installation path of custom nodes, but will add an extra path search when starting ComfyUI. You still need to complete the installation of custom node dependencies in the corresponding environment to ensure the integrity of the running environment.

Below is a simple configuration example (MacOS), please modify it according to your actual situation and add it to the corresponding configuration file, save it and restart ComfyUI for the changes to take effect:

my_custom_nodes:
  custom_nodes: /Users/your_username/Documents/extra_custom_nodes