Install, Update, and Uninstall on Windows#

Learn how to install, update, or remove AI Workbench on Windows.

Agree to the NVIDIA AI Product Agreement

Before downloading or installing, please read the NVIDIA AI Product Agreement, the NVIDIA AI Workbench Shared Security Model, and our Data Privacy Policy.

By downloading, installing, or using the NVIDIA AI Workbench software, you agree to the terms of the NVIDIA AI Product Agreement (EULA). If you do not agree to the terms of the EULA, you are not authorized to download, install, or use NVIDIA AI Workbench. Before downloading and using NVIDIA AI Workbench, please read our Shared Security Model.

Windows Prerequisites#

  • You Windows version must support WSL2

  • You must have administrator privileges and User Account Controls (UAC) enabled

  • System resources:

    • 1 GB of free disk space for the application files

    • 30-40 GB of disk space for containers

    • 16 GB of RAM

  • (Optional) An NVIDIA GPU with the latest NVIDIA GPU drivers

Important

Before you install AI Workbench you must run sufficient updates for Windows 10 or Windows 11 to get the build number high enough to support WSL2.

  • Windows 10 build must be 19045.3208 or higher.

  • Windows 11 build must be 22621 or higher.

Install AI Workbench on Windows#

  1. Log into the Windows user account that will use AI Workbench.

  2. Download the Desktop App Installer for Windows (AI Workbench Installer)

  3. Double click the installer file to launch the installation, and follow any prompts.

    • When requested, authorize Windows to allow installation steps (part of UAC)

    • If prompted, select a container runtime, either Docker Desktop or Podman.

  4. Follow the onboarding steps to clone the Tutorial Project and get started.

Update AI Workbench on Windows#

When updates are available, automated notifications will appear every time you open AI Workbench until the update is installed.

When you get an update notification, just click Update Now to begin the update and follow the prompts.

You can manually check for updates by right-clicking the AI Workbench icon in the Windows System Tray and selecting Check for Updates.

Note

If you update your local Workbench, you must also update any remote locations that are connected.

The connection requires API compatibility to work. See Update AI Workbench on a Remote System.

Uninstall AI Workbench on Windows#

  1. Quit AI Workbench completely by right-clicking the AI Workbench icon in the Windows System Tray and selecting Quit.

  2. Open a Windows Command Prompt, run the following command and follow the prompts:

    1wsl -d NVIDIA-Workbench -u root -- /home/workbench/.nvwb/bin/nvwb-cli uninstall --uid 1000 --gid 1000

  3. Search for and remove NVIDIA AI Workbench in the Windows Add or remove programs feature.

  4. Search on “workbench” in the AppData directory in the Windows system and remove the files. You may need to enable hidden files in Windows Explorer. See here for instructions.

  5. (Optional - Delete the WSL distribution) Open a Windows Command Prompt and run:

    wsl --unregister NVIDIA-Workbench

Warning

Removing the WSL2 distribution will permantely delete all files and data in the distribution. If you have any projects or data in the WSL2 distribution that you want to keep, make sure to back them up before uninstalling.

FAQs#

What Gets Installed and Where?#

Workbench puts files on both the Windows side and into the WSL2 distribution NVIDIA-Workbench.

Workbench installs no software dependencies on the Windows side other than Docker Desktop (if selected).

Windows Side: Everything installed here goes into program files or AppData.

  • C:\Program Files\NVIDIA AI Workbench: The Desktop Application and related files

  • C:\Users\<user-name>\AppData\Local

    • \NVIDIA Corporation\AI Workbench: Various config files and the credential-manager binary

    • \NVIDIA Corporation\AI Workbench Distro: The image file for the WSL2 distribution

    • \nvidia-ai-workbench-updater: The installer and updater files

  • C:\Users\<user-name>\AppData\Roaming\nvidia-ai-workbench: logs, local state file and a lock file

NVIDIA-Workbench: Workbench installs

  • Software dependencies: Git, Git-LFS, Podman (if selected) and NVIDIA Container Toolkit (if host has a GPU and Podman is selected)

  • Application files and binaries go into the folder /home/workbench/.nvwb

What if Docker Desktop is Already Installed?#

The installer will default to it and won’t ask you to select a container runtime.

How Do I Change to Podman Instead of Docker Desktop?#

If you have Docker Desktop installed, this is a little complicated. We don’t yet have a simple way to switch the runtimes. Basically, if Docker Desktop is installed it’s the default and Workbench won’t let you switch to Podman unless you uninstall Docker Desktop first.

  • Completely quit AI Workbench

  • Uninstall Docker Desktop from your Windows system

  • Go into the NVIDIA AI Workbench distro

  • Open the file /home/workbench/.nvwb/config.yaml in a text editor

  • Change the container runtime and buildtime fiels to podman instead of docker

  • Save the file

  • Restart AI Workbench and select the Podman container runtime when prompted

Do I Need to Install the NVIDIA GPU Drivers if my Windows System Has an NVIDIA GPU?#

Yes. You must install the NVIDIA GPU drivers on your own. We highly recommend using the NVIDIA App to install and manage your NVIDIA GPU drivers.

Why Do You Install a Dedicated WSL Distribution?#

Containers must run in a WSL distribution, so we use a dedicated distribution to avoid interfering with other WSL distributions you may have installed.

Next Steps#