€4/Month, Your Own Private AI Agent: Hermes Self-Hosting Guide

Why Self-Host an AI Agent?

Pasting code into ChatGPT to debug? That code might become training data. Letting an AI operate your server remotely? You’re handing over root access to a third party.

Hermes Agent (156K+ ⭐ on GitHub) is an open-source AI agent framework by Nous Research. Install it on your own VPS, and every conversation, memory, and skill file stays on your server — no third party ever touches it.

Open source + self-hosted = your data, your rules.

Cost Breakdown

A Hetzner CX22 (€3.99/mo, 2 vCPU / 4GB RAM / 40GB SSD) is plenty. Or go cheaper with RackNerd at $1.50/mo (1 core / 1GB).

API costs depend on usage. For daily coding, DeepSeek V3 ($0.27/M input tokens) runs about $20-30/mo with heavy use. You only pay for compute, not a subscription fee.

For reference: ChatGPT Plus is $20/mo (no self-hosting), Claude Pro is $20/mo (no self-hosting).

Deployment Steps

1. Get a VPS

Hetzner CX22 (€3.99/mo), Ubuntu 24.04. Oracle Cloud free ARM tier works too.

For users in Asia/Pacific: RackNerd $1.50/mo or BuyVM $3.50/mo are solid alternatives.

2. Install Dependencies

ssh root@your-vps-ip

apt update && apt upgrade -y
apt install -y curl git ffmpeg build-essential

ffmpeg is needed if you plan to generate video/audio via cron jobs.

3. Install Hermes Agent

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
source ~/.bashrc

The installer handles: uv (fast Python package manager), Python 3.11, Node.js, ripgrep (session search), and all Python dependencies. Takes about 30 seconds.

4. Verify

hermes doctor

All green? Good. Red errors are usually missing system packages — just apt install them.

5. Configure a Model

hermes model

Interactive picker, or set directly:

hermes config set provider openrouter
hermes config set openrouter_api_key sk-or-v1-your-key
hermes model openrouter/anthropic/claude-sonnet-4

Budget option: DeepSeek V3, Qwen 2.5, or Llama 3 — capable enough for coding, much cheaper.

6. Set Up the Telegram Gateway (Recommended)

Install it once, then control Hermes from your phone:

hermes gateway setup

Follow the prompts:

Open Telegram, search @BotFather
/newbot to create a bot, pick a name
Copy the Bot Token, paste it in

hermes gateway start

Now message your bot — Hermes responds instantly.

Keeping It Running 24/7

Option A: crontab (simplest)

crontab -e
# Add:
@reboot cd /root && hermes gateway start &

Option B: systemd service (more reliable)

# /etc/systemd/system/hermes-gateway.service
[Unit]
Description=Hermes Gateway
After=network.target

[Service]
ExecStart=/root/.hermes/hermes-agent/venv/bin/python -m hermes_cli.main gateway run
WorkingDirectory=/root
Restart=always
User=root

[Install]
WantedBy=multi-user.target

systemctl daemon-reload
systemctl enable --now hermes-gateway

Go with Option B. Restart=always means the process auto-recovers if it crashes — crontab can’t do that.

Post-Deployment FAQ

Gateway won’t start

Check ~/.hermes/logs/gateway.log. 90% of the time it’s a bad Bot Token — re-run hermes gateway setup.

Not enough RAM

On a 1GB VPS, Hermes + browser tools can OOM. Disable what you don’t need:

hermes tools disable browser
hermes tools disable vision

With only terminal / file / web enabled, memory usage drops below 200MB.

ARM architecture (Oracle free tier)

The installer handles ARM compatibility automatically. Some Python packages may need compilation — initial setup takes a bit longer.

Model API timeouts

Free-tier OpenRouter models time out frequently. Switch to a paid model or bump the timeout:

hermes config set agent.max_turns 120

How to update

hermes update

Then /restart or restart the gateway.

Disk Management

Hermes saves session logs by default. Over time this can grow to several GB:

hermes sessions prune --older-than 30

Or set auto-cleanup in config.yaml:

sessions:
  retention_days: 30

What’s Next

You now have a 24/7 AI agent that respects your privacy. What can it actually do for you?

This is Part 1 of a series. Coming up:

Part 2: Auto-publishing blog posts with Hermes — from writing to Git push
Part 3: Server monitoring with Hermes — alerts, log analysis, auto-remediation
Part 4: Running a Telegram channel with Hermes — scheduled posts, trending curation, rewriting
Part 5: Data ETL pipelines with Hermes — collect, clean, store

The core idea: The most valuable use of an AI agent isn’t chatting in the cloud — it’s working on your own infrastructure.

See you in Part 2.

Series index: Part 2: Auto-publishing blog posts with Hermes (coming soon)