ComfyStream is available as a docker image at livepeer/comfystream:latest. An NVIDIA GPU is required to run ComfyStream.

Choose a platform to install ComfyStream:

Local GPU

Run ComfyStream and ComfyUI on Linux or Windows with a local GPU

Remote GPU

Run ComfyStream and ComfyUI on RunPod with a rented GPU

​
Local GPU

​
Prerequisites

First, install the required system software:

​
Setting Up Local Environment

  1. Create directories for models and output:
mkdir -p ~/models/ComfyUI--models ~/models/ComfyUI--output
  1. Pull and run the container:
docker pull livepeer/comfystream:latest

If using Windows, ensure Docker Desktop is running first

docker run -it --gpus all \
-p 8188:8188 \
-p 8889:8889 \
-p 5678:5678 \
-v ~/models/ComfyUI--models:/workspace/ComfyUI/models \
-v ~/models/ComfyUI--output:/workspace/ComfyUI/output \
livepeer/comfystream:latest --download-models --build-engines --server

The --download-models and --build-engines flags will download required models and build TensorRT engines. This process may take some time depending on your network connection and GPU hardware and is only needed on the first run or when adding new models.

​
Remote GPU

​
Deploy on RunPod

  1. Deploy using the RunPod template livepeer-comfystream

First-time deployment to a network volume takes approximately 20-45 minutes depending on pod performance. For faster deployment without data persistence, use the RunPod template livepeer-comfystream-novolume

  1. When using the livepeer-comfystream template, create and select a network volume to persist files.
  2. Select an appropriate GPU (ex: RTX 4090) and click Deploy On-Demand
  3. Click the Logs button to monitor deployment progress.
  4. Once the container is fully running, you can access ComfyUI by clicking Connect

​
Accessing ComfyUI

​
Accessing ComfyStream

  1. From ComfyUI, click the ComfyStream menu button:

  2. Click Server Settings to verify ComfyStream is configured to bind to the correct interface and port

    When deploying to remote environments like RunPod, you will need to set the Host to 0.0.0.0.

  3. Click Save.

  4. Open the ComfyStream menu again, then click Start ComfyStream Server. Wait for the server status indicator to turn green.

You can also monitor ComfyStream server logs in ComfyUI’s log terminal tab.

  1. Click Open ComfyStream UI to launch in a new tab. Alternatively, you can double-click the node graph and search for ComfyStream UI Preview