AI agent hosting

AI Agent Hosting — Run Hermes Agent 24/7 on Your Own VPS

A self-hosted AI agent that stays on when your laptop is closed. Hermes Agent comes pre-installed on a Cloud VPS, keeps its memory between conversations, and answers you from Telegram, Discord or Slack. Start with a free Nous Portal account and no API key, or bring your own key. Your server, root access.

24/7An always-on service, not a terminal tab
MemoryPersistent, stored on your own disk
No API keyNeeded with Quick Setup; or use your own OpenRouter, Anthropic or OpenAI key
RootSSH access, Docker for extra tools
Works with: Hermes Agent · Claude Code · Nous Portal · OpenRouter · Anthropic · OpenAI — Hermes Agent is pre-installed as a service, Claude Code is an optional one-command install, Nous Portal needs only a free account, and the other providers are used with your own API key.

Why a VPS for your agent

An agent is only useful while it is running

Always on

An agent on your laptop stops when the lid closes. On an AI agent VPS it runs as a service, keeps its schedule and answers at three in the morning.

Your data on your server

Conversations, memory and files stay on a disk only you can log in to, in a France datacenter, not inside someone else's hosted agent product.

Your account or your keys

Sign in with a free Nous Portal account, or bring a key from OpenRouter, Anthropic or OpenAI and switch models whenever you like. With your own key you pay the provider directly, with no markup from us.

Full root access

Install tools the agent can call, add a database, run a second agent or the Claude Code CLI next to it. It is a normal Linux server.

How it works

From order to the first message

  1. 1

    Order a size

    Pick a plan and a billing cycle. The server is delivered with the agent stack already on it, and the credentials land in your client area and inbox.

  2. 2

    Log in and run ai-agent-setup

    Choose Quick Setup to sign in with a free Nous Portal account (no API key needed), or Full setup to paste your own OpenRouter, Anthropic or OpenAI key. A reminder appears at login until setup is done.

  3. 3

    Connect your Telegram bot and start chatting

    Give the setup a bot token for Telegram, Discord or Slack and your own user ID, then talk to your agent from your phone. This step is optional and can be done later.

What the wizard asks for

Quick Setup: a sign-in with your free Nous Portal account, no API key needed. Full setup: your own OpenRouter, Anthropic or OpenAI key. Optionally a bot token and your own user ID for Telegram, Discord or Slack, and whether to install the Claude Code CLI. The wizard does not start by itself; a reminder to run ai-agent-setup appears at login.

What you bring

A free Nous Portal account for Quick Setup, where no API key is needed, or your own account with a model provider for Full setup. For chat, a bot you created in the app you want to use. None of these is supplied by us, and no API credits are included.

Changing your mind

Keys, models and bot tokens can be changed after setup by running ai-agent-setup again. Updating the agent is one command, hermes update, and the service restarts automatically.

What's pre-installed

A self-hosted AI agent, ready when you log in

Hermes Agent hosting without the install guide: the pieces are in place, and nothing runs until you have completed the setup. Start with a free Nous Portal account and no API key, or use your own OpenRouter, Anthropic or OpenAI key, and add Claude Code from the wizard if you want it.

Hermes AgentThe open-source agent from Nous Research, installed natively and running as an always-on service under its own agent user, with persistent memory on your disk.
Setup wizardRun ai-agent-setup after you log in; a reminder appears at login. Quick Setup signs in with a free Nous Portal account, Full setup takes your own API key.
DockerDocker Engine and Compose are installed for the extra tools you add next to the agent: a database, a browser, a second service.
Optional Claude Code installerOne command from the wizard installs the Claude Code CLI. You sign in with your own Anthropic account; nothing is bundled.
Model providers: Nous Portal · OpenRouter · Anthropic · OpenAI
Chat channels: Telegram · Discord · Slack

AI agent VPS plans

The Cloud VPS sizes, with the agent already on them

Per month, incl. VAT, billed monthly or annually with two months free. No setup fee, cancel any time. Model usage is billed by your provider, not by us.

AI Agent VPS M

Agent plus its own services

vCPU
4 vCores
Memory
8 GB
Storage
75 GB NVMe
Port
1 Gbps
Traffic
Unlimited
IP
1 IPv4 + 1 IPv6
Backup
Daily snapshot
Agent
Hermes Agent as a service
GPU
None
Management
Self-managed

Room for a database, a browser tool or a second agent in Docker alongside Hermes Agent.

$15.99 /mo inc. VAT
$159.90/year
Buy now

AI Agent VPS L

Several agents or heavier tools

vCPU
6 vCores
Memory
12 GB
Storage
100 GB NVMe
Port
1 Gbps
Traffic
Unlimited
IP
1 IPv4 + 1 IPv6
Backup
Daily snapshot
Agent
Hermes Agent as a service
GPU
None
Management
Self-managed

Multiple agents, code execution sandboxes or builds, with memory to spare. Still no GPU.

$22.99 /mo inc. VAT
$229.90/year
Buy now
NVMe storage and a 1 Gbps port
Unlimited traffic and Anti-DDoS
Daily snapshot, restorable by you
Same price as a plain Cloud VPS of the same size
30-day money-back guarantee. No minimum term: cancel any time. Refund policy

Honest specs

What this server is, and what it is not

Read this before you order. It saves both of us a ticket.

  • No GPU. These are CPU servers. They host the agent, not the model.
  • Models run through an API. Quick Setup uses a free Nous Portal account and needs no API key. With Full setup you use your own OpenRouter, Anthropic or OpenAI key and pay that provider for usage.
  • Local models are a small-model option. Ollama will run on CPU, but only small models are practical and responses are slow.
  • Outbound SMTP ports 25, 465 and 587 are blocked by default. They are opened on request for legitimate use. Sending through a mail provider's HTTP API works without any change.
  • Acceptable use applies to automated agents. What your agent does on the network is your responsibility. Acceptable use policy
  • Self-managed. Operating system updates, the firewall, SSH hardening and updating the agent are your side of the deal. We look after the platform, the network and reinstalls.

FAQ

Before you order

Do I need a GPU?
No, and these servers do not have one. Hermes Agent sends its prompts to a model over an API; the VPS runs the agent itself, its memory and its tools, which is CPU and RAM work. Running a model locally with Ollama is possible, but on CPU it is only practical for small models and it is slow.
Which models can I use?
With Quick Setup you sign in with a free Nous Portal account and no API key is needed. With Full setup you paste your own OpenRouter, Anthropic or OpenAI key and can use any model that provider offers through its API. You can switch later by running the setup again. With your own key you pay the provider directly; no API credits are included with the server.
Is Claude Code included?
No. The setup offers an optional one-command install of the Claude Code CLI. You sign in with your own Anthropic account, and its usage is billed by Anthropic under your account.
Can I run OpenClaw or other agents?
Yes. You have root, and Docker is installed for extra tools, so any agent that runs on Linux will install. Only Hermes Agent is set up for you; anything else you install and maintain yourself. Support covers the server, not the configuration of other software.
Where is my data stored?
The agent's memory, its configuration and any API keys you enter are stored on your own VPS, in our France datacenter. What the agent sends to a model goes to the provider you chose and is handled under that provider's terms.
Who can talk to my agent?
Only the chat accounts you allow. During setup you enter your own Telegram, Discord or Slack user ID, and the agent answers only that account. Keep the bot private, because the agent can run commands on your server.
Can I upgrade later?
Yes. Open a ticket and we move you to the next size; the disk is kept and the change is a short reboot.
What happens if my agent misbehaves?
You are responsible for what your agent does, exactly as if you had done it by hand. Our acceptable use policy applies to automated agents: no spam, no scanning or attacking other systems, no abuse of third-party services. Outbound SMTP ports 25, 465 and 587 are blocked by default. If we receive a report we contact you first; for serious or ongoing abuse we suspend the server and tell you afterwards.
What is the refund policy?
30-day money-back guarantee. No minimum term: cancel any time. The refund policy on our legal pages has the full terms.

Give your agent a server of its own

Order a size, run ai-agent-setup and message your Hermes agent from your phone. No API key is needed with Quick Setup. From $8.49 a month, cancel any time.

Deploy your agent

n8n Hosting

Workflows rather than a chat agent: n8n pre-installed in Docker on the same sizes, with unlimited executions.

Cloud VPS

The same servers with a plain Linux install, for when you would rather build the stack yourself.

Managed VPS

We set up, patch, back up and monitor the server for you. For sites and applications, not for this agent image.