Hermes Agent, without the guesswork

Hermes Agent is an open-source AI agent from Nous Research that runs on hardware you control. Unlike a hosted assistant, it keeps its memory on your own disk, connects to your own messaging apps, and costs nothing to license. Unlike a chatbot, it remembers — and gets more useful the longer you run it.

It also ships a release most weeks, and not all of them are quiet. This site exists for the part the official docs do not cover: what actually changed in the latest version, what it broke, and the fix that actually works. Every article says which release it was tested against, and gets updated when a new one invalidates it.

Latest tracked release: v0.20.0 — The "Herald" release — Hermes gets a voice you can interrupt, talks to other agents over A2A v1.0, and cites its sources. Rolls up everything since v0.19.0, including the v0.19.1 patch window. All releases →

Hermes Agent at a glance

Built byNous Research
LicenseMIT — free, including commercial use
Runs onLinux, macOS, Windows, WSL2
InstallDesktop installer, pip install hermes-agent, or the official install script
HardwareNo GPU needed with a hosted model provider; 4 GB RAM is the practical floor
MemoryPersistent across sessions, stored locally
InterfacesCLI, TUI, desktop GUI, plus chat gateways for Telegram, Discord, Slack, WhatsApp, Signal and more
DataStays on your machine — no telemetry, no cloud lock-in
Current versionv0.20.0

Start here

Pick whichever describes you right now.

Fixes

Concrete solutions to Hermes Agent errors, ranked by how often they hit.

Guides

Install and integration walkthroughs for specific platforms and providers.

Common questions

What is Hermes Agent?

Hermes Agent is an open-source, self-hosted AI agent framework from Nous Research, released under the MIT license. It runs on your own machine, keeps memory that persists across sessions, and connects to messaging platforms like Telegram, Discord and Slack through a gateway process.

Is Hermes Agent free?

The software itself is free and MIT-licensed. What costs money is whatever model provider you point it at, plus a server if you want it running around the clock. For most people the model usage is the larger bill by a wide margin.

What makes Hermes different from a normal chatbot?

Persistent memory. A chatbot starts each session blank; Hermes keeps what it learned on disk you control and can search it later. It also turns successful workflows into reusable skills, so the same task gets cheaper the second time.

Do I need a GPU to run Hermes Agent?

Not if you use a hosted model provider, which is how most people run it. Hermes orchestrates; the model runs elsewhere. A GPU is only needed if you also intend to serve the model locally, which is a completely different budget.

How often does Hermes Agent get updated?

Roughly weekly. That pace is a genuine advantage and a genuine hazard: fixes land fast, but upgrades can break a working setup. Read the changelog before pulling, and snapshot first.

Why trust this site

Everything here is tested against a real install before it goes up, and each article carries the version it was verified on plus a genuine last-updated date. When a release breaks something documented here, the article gets corrected rather than left to rot. This is an independent resource and is not affiliated with Nous Research — the official repo and its docs remain the source of truth for the software itself. More about how this works →