Is Hermes Agent open source?

Yes, Hermes Agent is released under the MIT license by Nous Research. You can freely self-host it on your own VPS.

What's the difference from a cloud AI assistant?

Unlike a proprietary cloud assistant, Hermes runs on your own infrastructure: its memory and learning stay within your VPS (`~/.hermes` volume), with no dollar subscription and no data entrusted to a third party.

What VPS configuration does Hermes Agent need?

Plan for 4 GB of RAM to build the image and comfortably run the agent and its dashboard. The project is known to be lightweight and works on small configurations once installed.

How do I preserve what the agent has learned?

All of Hermes's memory and skills are stored in the `~/.hermes` volume. Back up this directory (or mount it on a persistent volume) to retain the learning across redeployments.

Build, host and run AI solutions.

Hermes Agent

The open-source AI agent from Nous Research that learns from you — self-hosted, MIT, sovereign.

4 GB RAM 2 vCPU Available

Source code

Tech stack

DockerPython

Minimum RAM4 GB

Minimum CPU2 vCPU

Compatible OSubuntu-24.04

Hermes Agent is the open-source AI agent from Nous Research, launched in early 2026 and now one of the fastest-growing agent projects on GitHub. Its distinctive trait: it "grows with you" — it creates skills from experience, refines them through use, searches its own past conversations and builds a model of who you are across sessions.

Under the MIT license, Hermes Agent self-hosts on a simple VPS. All of its memory and learning live within your infrastructure (`~/.hermes` volume), and you connect the LLM provider of your choice. It's a sovereign alternative to proprietary cloud AI agents: no dollar subscription, no memory of your exchanges stored with a third party.

Key features

Self-learning agent: creates and refines its skills from experience

Persistent memory: remembers your exchanges from one session to the next

Searches its past conversations to respond better

LLM provider of your choice, local models possible

Lightweight: runs on an entry-level VPS

Open source (MIT), by Nous Research

When to use this solution?

1

An assistant that knows you

An agent that retains your preferences, your projects and your context over the weeks, instead of starting from scratch with each conversation.

2

Automation that improves

Hermes learns from every completed task and builds reusable skills, gaining autonomy over time.

3

Sovereign and private AI

Memory and learning stay on your VPS; no agent data is entrusted to a proprietary cloud service.

Deploy Hermes Agent on your VPS

Guide optimized for ServOrbit Cloud VPS.

01

Order the VPS

Choose a ServOrbit Cloud VPS (4 GB RAM recommended for the build and execution) running Ubuntu 22.04.

02

Install from source

Hermes Agent is built from the official repository: the script clones the project, builds the Docker image and launches the stack (docker compose up -d).

03

Configure the LLM provider

Fill in the .env file (copied from .env.example) with the provider and the key of the model you want to use.

04

Access the dashboard

The Hermes dashboard runs locally on the VPS; place an Nginx or Caddy reverse proxy with SSL in front to access it from your domain over HTTPS.

05

Let the agent learn

Over the course of your exchanges, Hermes enriches its memory and its skills (~/.hermes volume) — back up this volume to preserve the learning.

Frequently asked questions

Hermes Agent is an open-source (MIT) AI agent developed by Nous Research. Self-hostable, it stands out for its continuous learning: it creates skills from experience and remembers your exchanges from one session to the next.

Complementary solutions

Artificial Intelligence

Activate Hermes Agent on your infrastructure.

Dedicated Cloud VPS — IPv4 included, European datacenter, support included. Your data never leaves your server.

Recommended configuration: 4 GB RAM · 2 vCPU