Installation
The OpenFork desktop client is available for Windows and Linux. It manages Docker containers for every AI workflow, so your only requirement is an NVIDIA GPU and the installer.
NVIDIA GPU required. AMD and Apple Silicon are not supported. Minimum 8 GB VRAM recommended for most models.
Windows
The Windows installer is fully automated — it sets up the entire environment including WSL2, Docker, and the GPU worker without any manual configuration.
1
Go to the Download page and download the Windows installer (.exe).
2
Run the installer. It will automatically:
- ›Download and configure an Ubuntu WSL2 distro
- ›Install Docker inside that distro
- ›Configure the NVIDIA Container Toolkit for GPU passthrough
- ›Install and set up the OpenFork Python client
3
Once installation completes, open OpenFork Desktop from the Start menu.
4
Sign in with the same Google account you use on openfork.video.
5
Go to the Docker tab, find a model image you want, and click Download. The container will pull in the background.
6
Once the image is ready, click Start Client. Your GPU is now connected to the network and ready to process jobs.
Model images are large — 80 GB to 220 GB depending on the model. They download once. To free disk space, delete the Docker image from the Docker tab.
Linux
On Linux, Docker and the NVIDIA Container Toolkit must already be installed. The desktop app ships as an AppImage — a self-contained executable that needs no package installation.
Prerequisites
Before running OpenFork, ensure the following are installed and working:
- ›Docker Engine — verify with
docker run hello-world - ›NVIDIA Container Toolkit — verify with
docker run --gpus all nvidia/cuda:12.0-base nvidia-smi - ›NVIDIA driver (latest stable) — verify with
nvidia-smi
Install the desktop app
1
Go to the Download page and download the Linux AppImage.
2
Make it executable. Either right-click the file → Properties → Executable, or run:
chmod +x Openfork-Client-linux-x86_64.AppImage
3
Launch the app:
./Openfork-Client-linux-x86_64.AppImage
4
Sign in with your Google account and follow the same steps as Windows — download a model image from the Docker tab, then start the client.
On some Linux distributions you may need to install
libfuse2 for AppImages to run: sudo apt install libfuse2Headless / Python client only
If you don't need the desktop UI, you can run the Python client directly from the openfork_client repository:
git clone https://github.com/besch/openfork_client cd openfork_client python -m venv venv ./venv/bin/python -m pip install -r requirements.txt ./venv/bin/python cli.py --dgn-api-key <your_dgn_api_key> --service auto --community-mode all