Skip to content

How-ToCPU

How to Install AMD's AI Bundle: ComfyUI, Ollama, LM Studio and PyTorch in One Go

AMD's AI Bundle is an optional part of the Adrenalin driver installer that sets up PyTorch on Windows, ComfyUI, Ollama, LM Studio and Amuse in one go. AMD says it supports Radeon RX 9000 cards, RX 7700 and above from the RX 7000 series, and Ryzen AI 300, 400 and Max processors, and needs about 35GB for everything plus an internet connection.

By , Systems & Technical Editor

4 min read

How to Install AMD's AI Bundle: ComfyUI, Ollama, LM Studio and PyTorch in One Go

Setting up local AI on a Windows PC used to mean installing Python, frameworks and apps one by one, then hoping their versions worked together. AMD's AI Bundle aims to fix that for Radeon and Ryzen AI owners: it's an optional part of the AMD Software: Adrenalin Edition installer that sets up popular AI tools for you.

This guide is based on AMD's AI Bundle FAQ, AMD's blog post introducing the feature and the release notes for Adrenalin 26.11.1.

What's in the AI Bundle?

AMD says the AI Bundle arrived as an optional feature in Adrenalin 26.1.1. It can install:

  • PyTorch on Windows, including Visual Studio Code, Python 3.12.0 and a recent version of PyTorch for ROCm on Windows, so you can start writing AI code straight away

  • ComfyUI, the node-based app for image generation and other AI workflows

  • Ollama, for running large language models (LLMs) from the command line or other apps

  • LM Studio, a desktop app for downloading and chatting with LLMs

  • Amuse, an app for AI image generation

If you're new to these apps, our guides to installing ComfyUI, running models with Ollama and LM Studio and generating images with Amuse explain what each one does. We also compare ComfyUI vs Amuse.

Which hardware is supported?

According to AMD, the AI Bundle supports:

  • AMD Radeon RX 9000 series graphics cards

  • AMD Radeon RX 7000 series graphics cards, from the RX 7700 series upwards

  • AMD Ryzen AI Max series processors

  • AMD Ryzen AI 400 series processors

  • AMD Ryzen AI 300 series processors

If you have an older Radeon card, or an RX 7600, you can still install most of these apps manually. Our AMD Radeon local AI guide covers that.

What you need before you start

  • Disk space: AMD says the full AI Bundle needs about 35GB of free space. That's before you download any AI models, which can be several gigabytes each.

  • An internet connection: AMD says this is required, because some components come from third-party developers' websites.

  • The latest Adrenalin driver. At the time of writing, that was Adrenalin 26.11.1, which also fixes excessive writes to the system drive on RX 7000 series cards and newer, as covered in our news report on the SSD writes fix.

Step 1: Download the Adrenalin installer

Download AMD Software: Adrenalin Edition from AMD's website, or use the update option in the Adrenalin app if it's already installed. Our graphics driver update guide covers both methods.

Step 2: Choose the AI Bundle option

AMD's blog says that during the Adrenalin install, you simply tick the AI Bundle option. AMD's FAQ adds that you can see the file size of each app with the Custom Install option or the AMD Installer Manager during setup.

You don't have to install everything. If you only want to chat with LLMs, for example, choose LM Studio or Ollama and skip PyTorch and ComfyUI to save space.

Step 3: Finish the install and launch the apps

Let the installer download and set up the components you chose. When it finishes, launch the apps from the Start menu. For PyTorch, AMD says to open Visual Studio Code with the global Python 3.12.0 environment to start working with PyTorch ROCm on Windows.

Then download a model to try, choosing one that fits in your graphics memory. Our VRAM usage guide shows how much is free, and our VRAM explainer covers how much different AI tasks need.

If the install fails

AMD's 26.11.1 release notes list a known issue: AI Bundle components may fail to install in some regions with limited access to HuggingFace and GitHub. If you have problems, check that your connection, firewall or VPN isn't blocking those sites, then run the installer again.

How to remove AI Bundle apps

AMD says you can uninstall the apps individually through Windows' Installed apps list, or Programs and Features in Control Panel. ComfyUI and the PyTorch libraries need a manual clean-up:

  • ComfyUI: in File Explorer, turn on View > Show > Hidden items, go to your user folder's AppData location that AMD's FAQ lists, and delete the ComfyUI folder.

  • PyTorch: with hidden items shown, go to C:\Users\[your user name]\AppData\Local\Programs\Python\ and delete the Python312 folder.

Before deleting the Python312 folder, make sure no other software you use depends on that Python installation.

Our verdict

AMD's AI Bundle is the easiest way to set up local AI on a recent Radeon graphics card or Ryzen AI PC. Tick the AI Bundle option during the Adrenalin install and you get PyTorch, ComfyUI, Ollama, LM Studio and Amuse in one go. It needs an RX 9000 card, an RX 7700 or better from the RX 7000 series, or a Ryzen AI 300, 400 or Max processor, plus about 35GB of disk space for everything. Use Custom Install to pick only the apps you need.

The bottom line

AMD's AI Bundle is the easiest way to set up local AI on a recent Radeon graphics card or Ryzen AI PC. Tick the AI Bundle option during the Adrenalin install and you get PyTorch, ComfyUI, Ollama, LM Studio and Amuse in one go. It needs an RX 9000 card, an RX 7700 or better from the RX 7000 series, or a Ryzen AI 300, 400 or Max processor, plus about 35GB of disk space for everything. Use Custom Install to pick only the apps you need.

Sources

  1. 01AMD, AI Bundle frequently asked questions
  2. 02AMD, AMD Software: Adrenalin Edition AI Bundle blog
  3. 03AMD, Adrenalin Edition 26.11.1 release notes

About the author

Marcus Thorne

Systems & Technical Editor

Marcus writes Forge PC's step-by-step build, troubleshooting and system optimisation guides. His background spans Windows kernel and power-state diagnostics, UEFI/BIOS configuration, NVMe storage arrays, ATX 3.1 power delivery and running local AI and LLM workloads on consumer graphics cards.

More from Marcus Thorne

Read next

How to Run a Local LLM on an AMD Ryzen AI NPU With FastFlowLM

How-To

How to Run a Local LLM on an AMD Ryzen AI NPU With FastFlowLM

FastFlowLM runs language models on the XDNA 2 NPU in Ryzen AI chips. Update the NPU driver to 32.0.203.311 or later, install flm-setup.msi, test with flm run llama3.2:1b and use flm serve for an OpenAI-compatible local API.

Marcus Thorne3 min