Hermes never forgets
Talk to it on Telegram, pick up on Discord, check in from iMessage.
24+ platforms, one memory. It never forgets.
Starting at €5.00/mo
What is Hermes Agent?
Hermes Agent is a free, open-source personal AI assistant built by Nous Research. You talk to it on Telegram, WhatsApp, Discord, Slack, and 24+ other platforms, and it keeps a single persistent memory across all of them. It is model-agnostic: pick from 400+ models through Nous Portal or OpenRouter, or plug in your own provider keys. Because the models run in the cloud, the agent itself is light and runs around the clock on a small VPS.
213k+
GitHub stars
24+
Messaging platforms
400+
AI models
v0.18
Updated weekly
Figures checked July 2026. Official Hermes Agent documentation
Why run Hermes Agent on a VPS?
An agent with memory needs a server that stays on.
Persistent memory needs a persistent server
Your agent learns your preferences and remembers across sessions. A VPS keeps that knowledge alive 24/7. No resets, no starting over.
One agent, 24+ platforms
Telegram, Discord, Slack, WhatsApp, Signal, iMessage, WeChat, Android, and more. Ask a question on your phone, get the answer on your laptop. Same agent, same memory, wherever you are.
Terminal, dashboard, or both
Hermes Agent runs in a real terminal with multiline editing and autocomplete, plus a local web dashboard to manage config, sessions, and skills. A VPS gives it a proper Linux server that stays on.
You control what leaves
Your code and credentials stay on your European VPS. You pick which model provider gets your prompts, and how much context you share.
How much VPS does Hermes Agent need?
Hermes Agent runs comfortably on a 2 GB VPS. The CLI alone uses roughly 200 to 350 MB of RAM; with the messaging gateway running, plan for 400 to 950 MB plus headroom for tools and the OS. Models run in the cloud through your API key, so you never need a GPU.
| Your setup | RAM | Plan |
|---|---|---|
| Terminal only, trying it out | 1–2 GB | Starter |
| Daily driver, one to three messaging channels | 2 GB | Starter (recommended) |
| Several channels, skills, browser tool | 4 GB | Pro |
| Many channels, large memory stores, heavy tooling | 8 GB | Power |
RAM figures are community measurements of Hermes Agent without local models. Running local models needs far more resources.
Pick your Hermes Agent VPS
All plans include full root SSH access, instant deployment, and DDoS protection.
Power
Heavy workloads. Large memory store.
€0.0070/hr
- 1 vCPU
- 1 GB RAM
- 20 GB SSD
- 1 Gbps
Starter
Single channel. Light memory usage.
€0.0167/hr
- 1 vCPU
- 2 GB RAM
- 40 GB SSD
- 1 Gbps
Pro
Multi-channel. Persistent memory.
€0.0333/hr
- 2 vCPU
- 4 GB RAM
- 80 GB SSD
- 1 Gbps
Need a larger plan? View all Linux VPS plans
One command, ready to go
Our installer handles all dependencies. Works on Debian, Ubuntu, and RHEL.
curl -fsSL https://virtua.sh/i/hermes-ssh | bash Prefer to install step by step?
Full manual setup if you want more control.
Show manual steps
Deploy your VPS
Choose Debian 12 or Ubuntu 22.04 at checkout. Your VPS is activated instantly.
Deploy nowSSH and install Hermes Agent
Connect to your server and run the official one-line installer. It handles all dependencies automatically.
ssh root@your.server.ip curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash source ~/.bashrc Configure your model and API key
Run the setup wizard to pick your LLM provider and configure your API key.
hermes doctor hermes model Start chatting
Talk to your agent from the terminal, open the web dashboard, or launch the gateway to connect all your messaging platforms.
hermes chat hermes dashboard hermes gateway Pro Tip
Run hermes setup to configure model, tools, and messaging channels all at once.
Multi-channel ready
Connect Telegram, Discord, Slack, iMessage, WeChat, and 19 more platforms at once. One agent, shared memory.
Run 24/7
Use hermes gateway install to set up a systemd service that starts automatically on boot.
Run your agent safely
An always-on agent holds your API keys and can run commands. Four habits keep it yours alone.
SSH keys and a firewall
Disable password login, allow only SSH through the firewall, and keep the system updated. Standard server hygiene covers most real-world attacks.
Keep the dashboard private
The web dashboard binds locally by default. Reach it through an SSH tunnel instead of exposing it to the internet. Exposed agent dashboards are the most common self-hosting mistake.
Scope your API keys
Give the agent its own provider key with a spending limit, never your main one. Rotate it if anything looks off. Keep command approvals on for anything destructive.
Give it its own server
Run the agent on a dedicated VPS, not on your laptop or next to production workloads. If an experiment goes sideways, you rebuild a server, not your life.
Want the full hardening checklist? Follow the step-by-step Hermes Agent tutorial
Always up to date
One command to update. Your memory, skills, and config are preserved.
hermes update Pulls the latest code, reinstalls dependencies, and migrates your config. Your conversations, memories, and learned skills stay exactly where they are.
Why developers trust Virtua.Cloud
Independent European hosting, owned infrastructure, human support.
Since 2015
Over a decade of European hosting
99.99% Uptime
Reliable infrastructure with SLA
French Company
GDPR, French data protection
Fair Pricing
No hidden fees, hourly billing
SSH Root Access
Your server, your rules
Frequently Asked Questions
Common questions about Hermes Agent on a VPS.
Not sure yet? Talk to us, we run these too
Coming from OpenClaw?
We run both agents and wrote down what actually differs: footprint, memory design, security record, and day-to-day maintenance. Plus how to migrate in an afternoon.
Run Hermes Agent 24/7
24+ platforms, persistent memory, web dashboard. All in one place.