Run Hermes Agent Locally On AMD Ryzen™ AI Max+ Processors and Radeon™ GPUs

Hermes Agent by Nous Research is a major step forward for local, autonomous, AI agents. It is an incredibly powerful, fully autonomous AI agent with some serious capability behind it. It is optimized for local models and can take goals, break them down, work through steps, and drive execution in a way that feels advanced, capable, and undeniably next-generation. Hermes is what happens when autonomous AI moves beyond theory and starts delivering real-world impact.

AMD Agent Computers are designed for exactly this moment. This new category of always-on systems are built to run AI agents continuously across apps and tasks, so users can delegate work and let agents execute autonomously. This model depends on hardware capable of running AI locally and performantly. Systems powered by AMD Ryzen™ AI Max+ processors or Radeon™ AI PRO graphics are especially well suited for this category, with the horsepower for sophisticated local models and the architecture to support the parallel, multi-agent workloads that define agentic computing.

Running it locally on AMD Ryzen™ AI Max+ and Radeon™ AI PRO is fairly straightforward and grants you the control and immediacy that local execution makes possible. Hermes Agent brings the autonomy, AMD brings the hardware capability and LM Studio provides the path to get it running. Together, they create exactly what local AI should look like: fast, capable, privacy-focused, and full of potential. This is a powerful example of fully autonomous AI on AMD, and allows you to turn intent into action and start delegating your workflows to agents - all the while staying inside the familiar Windows ecosystem. 

How to setup and run Hermes Agent on AMD Ryzen™ AI Max+ processors or Radeon™ graphics

This guide walks through the setup on AMD Ryzen™ AI Max+ processors and Radeon™ AI PRO graphics using WSL2 (Windows Subsystem for Linux) combined with LLMs served over LAN through LM Studio (to easily traverse the WSL2 boundary).  Once set up, the conclusion is hard to miss: Hermes Agent is incredibly capable, local AI is advancing quickly and AMD builds for this.

Installing and configuring LM Studio

1. Download LM Studio 
2. Start LM Studio and click on "Skip for now" for the model download prompt. 
3. Check "Developer Mode" and "Start LLM service on login"
4. Click on the "Model Search" Icon represented by a Robot and a Magnifying glass
5. Select "Qwen3.6 35B A3B" on the left hand side and click download on the right hand side. Wait for it to finish downloading. 

6. Click Ctrl + L to bring up model selection. Check "Manually choose load model parameters" and click on the LLM to bring up the load options. Check "remember settings" and check "show advanced settings".
7. Set the following settings: Context: 190000, GPU Offload: Max, Uncheck "Try mmap", Flash Attention: On, Remember settings: On. 

8. Press Ctrl + 2 to take you to the developer tab. 
9. Press the toggle infront of "Status: Stopped" to start the server. 
10. Click on Server Settings and set: Allow "Serve on Local Network". You will be prompted for permissions at this point, Uncheck "Auto unload unused JIT loaded models" and "Only keep last JIT loaded model". Click anywhere to close the context window. 
11. Click on Load Model. Select Qwen 35B A3B. The settings that we configured above should automatically populate. Click Load. 
12. The model should get loaded and you will see the status "READY". 
13. Keep this window open and move on. 

Installing WSL2 and Hermes Agent

14. Open Powershell as Administrator and install WSL and the distro (you might be prompted for a restart). If you have already installed a distro and want to start with a clean install you can run "wsl --unregister ubuntu-24.04":

15. Set a user and password and then run (you might be prompted for the password (sudo) and asked to continue):

16. Select quick setup and select "custom endpoint" as the provider. 

17. Copy paste the local server address from LM Studio (under API usage) and append "/v1" to the end of it. 

18. For the key just input "lmstudio". 
19. Leave context length blank as that will be automatically detected. 
20. Give it a name like " LM Studio". 
21. Setup any messaging platform if needed and you are all done!

From here, Hermes Agent can expand into live research, web-aware execution, and workflows that depend on current external information. Adding search capability gives the agent access to the live web, which is especially valuable for research, fact-finding, and tasks that require up-to-date context. A practical way to enable that is with a provider such as the Brave Search API (once you have the key - just ask Hermes to guide you through the setup).

For deeper setup guidance and next steps, start with the official Hermes Agent documentation.