Learning how to set up Hermes Agent is far easier than most people think, and I can get it running in one click in about 20 seconds.
I know what's going through your head right now.
You've seen the terminal commands, the GitHub repos and the configuration files, and you've decided this stuff isn't for you.
That fear puts a lot of people off, and I get it.
The good news is that it doesn't have to be like that anymore.
In this guide I'm going to walk you through the three ways I use to get Hermes running in one click.
The first is a free cloud option, the second is a free local option that runs on your own computer, and the third is a hosted option with zero technical setup.
By the end you'll know exactly which one fits your machine, and you'll have a working AI agent instead of a bookmarked tutorial.
The quick answer: how to set up Hermes Agent in one click
If you only read one section, read this one.
Here are the three routes side by side so you can pick in five seconds.
| Route | What it costs | What you need | Best for |
|---|---|---|---|
| Ollama + cloud model | Free, with token limits | Ollama installed and running | Most beginners and anyone on a normal laptop |
| Ollama + local model (Gemma 4) | Free, runs on your machine | A reasonably powerful computer and time to download the model | People who want a free API running locally |
| Max Hermes (hosted by MiniMax) | Paid, not free | A MiniMax account | People who never want to open a terminal |
My honest recommendation is to start with the Ollama route.
It's free, it takes one command, and you can swap models whenever something new comes out.
If you get stuck picking, pick Ollama and move on.
🔥 Want my full Hermes training, not just the install? Inside the AI Profit Boardroom, there's a full 2-hour Hermes course that takes you from setup to real automations. You also get four weekly coaching calls where you can share your screen and get your setup fixed, alongside 3,400+ members running AI agents. → Get access here
Why setting up an AI agent feels harder than it is
Most people never get past the install screen.
They open the GitHub page, see a wall of commands, and close the tab.
I've watched this happen with friends, clients and people in my community over and over again.
The problem isn't that Hermes is hard to use.
The problem is that the first five minutes look scary.
Once Hermes is running, you're just chatting to an agent that can actually take actions for you.
So the whole game is getting through those first five minutes as fast as possible.
That's exactly what the one-click routes below do.
Step 1: Install Ollama before you do anything else
The first thing you need is Ollama on your computer.
You can download it for free from ollama.com.
Ollama is the free tool that lets you run AI models, either through its cloud or locally on your own machine.
Install it like any normal app.
Then make sure Ollama is actually running before you move on to the next step.
This is the single most common mistake I see, so I'll say it twice.
Ollama has to be open and running, otherwise the one-click command has nothing to connect to.
Step 2: How to set up Hermes Agent with a free cloud model
This is the route I'd send most people down first.
Head over to the Models section on the Ollama website.
Pick whichever model you want to power Hermes.
If you're not sure what to choose, pick a lightweight model that's built for agentic tasks so it runs smoothly on your machine.
On the model page you'll see a command for launching it with Hermes.
Copy that command.
Open your terminal and paste it in.
Hit proceed when it asks you to confirm.
Wait a few seconds for everything to load.
That's it, and you now have Hermes Agent set up in one single click.
I'm not exaggerating when I say the whole thing takes about 20 seconds once Ollama is installed.
The one catch with cloud models
Cloud models on Ollama are free, but they come with token limits.
That means you can use them for real work, but heavy users will eventually hit a ceiling.
For testing Hermes, learning how it works and running everyday tasks, that ceiling usually isn't a problem.
If you start bumping into it regularly, that's your signal to look at the local route below.
Which cloud models I'd recommend with Hermes
These are the cloud models recommended for running Hermes through Ollama.
- Kimi K2.5 is one of the recommended cloud options.
- GLM 5.1 Cloud is another strong pick on the list.
- Qwen 3.5 Cloud works well with Hermes too.
- MiniMax M2.7 Cloud rounds out the recommended cloud models.
You don't need to agonise over this choice.
The beauty of the Ollama setup is that you can switch between models whenever you want.
When something new drops, you just launch it with Hermes and test it the same day.
Step 3: How to set up Hermes Agent locally with Gemma 4
The second free route runs the model on your own computer.
This means you've got a free API running locally, and Hermes talks to it directly.
The model I'd point you to is Gemma 4.
Gemma 4 is a free local model, and it comes in various sizes.
Some versions are super lightweight, and one of the ones I showed is only around 7GB.
That is very small for a local model.
The process is almost identical to the cloud route.
Make sure Ollama is running first.
Find the Gemma 4 model page on Ollama.
Copy the Hermes launch command for it and paste it into your terminal.
Hit proceed, and Hermes will launch with a local model instead of a cloud one.
From there you just use Hermes directly in your terminal as normal.
The honest trade-offs of running Hermes locally
Running locally sounds amazing, and it is, but there are two things you need to know.
The first is that it depends heavily on your hardware.
I run this on a Mac Studio, so it runs pretty smoothly for me.
If you've got an older laptop, a local model may feel slow or struggle.
The second is that local models take a while to download and install.
So here's the simple rule I use.
If you've got a powerful setup, go local.
If you don't have a powerful setup, stay on one of the cloud models.
Recommended local models for Hermes
The two local models recommended for Hermes are Gemma 4 and Qwen 3.6.
Start with Gemma 4 if your machine is modest, because the smaller versions are so light.
Bonus local option: NVIDIA Nemotron 3 Nano Omni
There's another local model worth knowing about if you like to tinker.
Ollama announced that NVIDIA's Nemotron 3 Nano Omni is available to run locally.
What makes it interesting is that it's designed for agentic tasks, and more importantly for really efficient tasks.
The main benefit is powering sub-agents.
So you could have a main "brain" model doing the thinking, and Nemotron 3 Nano Omni handling the smaller sub-agent jobs.
To run it with Hermes, you go to the Nemotron 3 page on Ollama, scroll down, and pick the Omni version.
Bear in mind that it's quite a big model, and you're looking at something like 28GB.
I'd also be honest that this setup gets more technical the more you customise it.
If you're brand new, park this one until you're comfortable with the basics.
Step 4: The no-terminal route with Max Hermes
Some people simply never want to see a terminal, and that's fine.
For you, there's Max Hermes.
Max Hermes is basically Hermes Agent hosted in the cloud by MiniMax, and you can start it in one click.
You can reach it through the MiniMax agent platform at agent.minimax.io.
I'll be honest that they don't make it the easiest thing to find.
Once you're there, you click Start Now.
Max Hermes wakes up, loads its skills, and in roughly 20 seconds it's running directly inside the chat.
You just talk to it in the browser like any other chat app.
What's good about Max Hermes
The biggest win is that it's completely non-technical.
You don't install anything, and you never touch a command line.
Another nice touch is that you can test MaxClaw and Max Hermes side by side on the same platform.
MaxClaw is OpenClaw, another AI agent, hosted in the cloud by MiniMax.
That lets you compare the two agents and see which one you prefer without installing either.
The other cool thing is that MiniMax can generate images and videos.
So when you use Max Hermes, those MiniMax powers work straight out of the box, without wiring up separate APIs.
What's not so good about Max Hermes
This route is not free, unlike the two Ollama methods.
You can't link it to Telegram or any other apps, so it lives only inside MiniMax.
You can't upload files to it either.
I've tested it, and my honest verdict is that it's not bad, but it's not great.
You also lose the customisation you'd get running Hermes in your own terminal.
That's the biggest difference between the hosted route and the Ollama routes.
If you just want a non-technical feel for what Hermes can do, it's a decent place to dip your toe in.
🔥 Not sure which Hermes route fits your setup? Inside the AI Profit Boardroom, you can post your exact machine and goals in the community and get help, or bring it to one of the four weekly coaching calls. The classroom has the full Hermes course plus a 6-hour OpenClaw course. → Join the Boardroom here
The cheat code: let Claude Code install Hermes for you
You can also go straight to the official Hermes GitHub page if you prefer the traditional install.
I know a lot of people struggle with GitHub if they're not used to it.
So here's what I'd do if I was non-technical.
Copy the GitHub link for Hermes.
Open Claude Code on your computer.
Paste the link in and simply say "set this up for me".
Claude Code will go off, read the instructions and install Hermes for you.
You can do exactly the same thing with Codex if that's what you use.
It's effectively another one-click setup, because all you do is hit enter and let the agent do the work.
I love this trick because it's the same move I use for almost every new tool now.
If you want to see how far you can take Hermes once it talks to other agents, read my breakdown of the Hermes MCP server setup with Codex.
What you get once Hermes Agent is running
Getting Hermes installed is only the start, so let's talk about what you actually have.
Hermes works with over 70 skills by default, which is a lot of capability before you've customised anything.
It runs with tons of different models, so you're never locked into one provider.
Because you've used Ollama, switching between models is trivial.
That matters more than people realise, because new models come out every week.
When one drops, you launch it with Hermes and test it in your own workflow the same afternoon.
That's a huge advantage over tools that lock you into a single model.
Once you're comfortable, one of the best things you can do is give your agent a memory of your business.
I've written up how I do that with notes in my Claude Obsidian setup guide, and the same thinking applies to Hermes.
Which Hermes setup should you pick?
Here's how I'd decide if I were starting from zero today.
| Your situation | The route I'd pick |
|---|---|
| You're a complete beginner on a normal laptop | Ollama with a free cloud model |
| You've got a powerful machine and want it free and local | Ollama with Gemma 4 or Qwen 3.6 |
| You want to experiment with sub-agents | Ollama with Nemotron 3 Nano Omni |
| You refuse to open a terminal and don't mind paying | Max Hermes on MiniMax |
| You want the official GitHub install but find it confusing | Paste the GitHub link into Claude Code or Codex |
If you're still torn, go with Ollama.
You can always try the others later, because none of these choices are permanent.
Common mistakes when you set up Hermes Agent
These are the issues I see most often in my community.
- People forget to start Ollama first, so the command has nothing to connect to.
- People pick a huge local model on a weak laptop and then decide Hermes is slow, when the hardware is the real bottleneck.
- People expect Max Hermes to connect to Telegram, but that hosted version only works inside MiniMax.
- People overthink the model choice, when switching later takes seconds.
- People treat the install as the finish line, when the real value comes from what you build afterwards.
Avoid those five and you'll be ahead of most people trying this for the first time.
FAQ: how to set up Hermes Agent
How do I set up Hermes Agent in one click?
Install Ollama from ollama.com and make sure it's running.
Go to the Models section, pick a model, and copy the Hermes launch command from that model's page.
Paste it into your terminal, hit proceed, and Hermes is running in about 20 seconds.
Is Hermes Agent free to set up?
Yes, two of the three routes are completely free.
The Ollama cloud route is free with token limits, and the Gemma 4 local route is free on your own hardware.
Only the hosted Max Hermes option on MiniMax costs money.
What's the best model to run Hermes Agent with?
For cloud, the recommended options are Kimi K2.5, GLM 5.1 Cloud, Qwen 3.5 Cloud and MiniMax M2.7 Cloud.
For local, the recommended options are Gemma 4 and Qwen 3.6.
You can switch any time, so don't let this decision slow you down.
Can I set up Hermes Agent without using the terminal?
Yes, Max Hermes on MiniMax starts with one click in your browser.
It's paid, it can't connect to Telegram or other apps, and it can't take file uploads.
Do I need a powerful computer to run Hermes Agent?
You only need a powerful computer for the local route.
The cloud route runs the model on Ollama's side, so a normal laptop is fine.
Can Claude Code install Hermes Agent for me?
Yes, paste the Hermes GitHub link into Claude Code and say "set this up for me".
It'll handle the install, and Codex can do exactly the same.
My verdict on the fastest Hermes setup
Hermes is one of the most capable free AI agents you can run today, and the install no longer needs to be the hard part.
Start with Ollama and a free cloud model, because it's the quickest win.
Move to Gemma 4 locally once you want a free API on your own machine.
Use Max Hermes only if the terminal genuinely stops you from starting.
Whichever route you pick, the important thing is that you have a working agent today rather than a perfect plan next month.
🔥 Ready to go from installed to automating real work? The AI Profit Boardroom has the full 2-hour Hermes course, new daily trainings in the classroom, a map to meet members in your city, and four weekly coaching calls. You'll be learning alongside 3,400+ members who run agents every day. → Get instant access
Related reading
- Hermes MCP server: how I connect Codex to Hermes so two agents work in parallel
- Claude Obsidian setup: how I turn a pile of notes into an AI brain
Also From Julian
- If you want my daily notes on AI agents and automation, browse the AI Profit Boardroom blog.
- For deeper step-by-step agent walkthroughs, I publish visual guides over at agentos.guide.
- If you're choosing which model to plug into Hermes, check my independent model scores on GoldieBench.
About Julian
I'm Julian Goldie, an AI entrepreneur, SEO expert, and founder of the AI Profit Boardroom (3,400+ members). I help business owners scale with AI agents, automation, and SEO.
- I've grown my YouTube channel to 400,000+ subscribers.
- I run a 7-figure AI agency, Goldie Agency.
- I share daily training inside the Boardroom.
- I've written multiple AI automation playbooks.
→ Get my best AI training inside the AI Profit Boardroom
📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉
🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉
Now you know how to set up Hermes Agent in one click, so pick your route and get your first agent running today.
