Hermes Agent VPS Setup: Complete 7-Step Guide to Self-Host Your AI Assistant

Self-hosting Hermes Agent on a VPS gives you a 24/7 AI assistant that never sleeps, remembers everything across sessions, and is reachable from your phone via Telegram, Discord, or WhatsApp — all for less than the cost of a Netflix subscription. If you have been running Hermes Agent on your desktop and want it available around the clock, or if you are an OFW who wants an AI assistant that stays online while you are at work, this Hermes Agent VPS guide walks you through every step from a fresh server to a fully deployed agent.

Key Takeaway

• A Hermes Agent VPS deployment costs $6.49 to $8.79 per month on a Hostinger KVM 1 or KVM 2 plan — cheaper than most AI subscriptions, with full control and no vendor lock-in.
• The entire setup takes under 30 minutes from a fresh Ubuntu VPS to a working Hermes Agent with messaging gateway, using the official one-line installer and systemd service.
• You do not need a GPU — the Hermes Agent VPS approach connects to cloud models (Nous Portal, OpenRouter, DeepSeek, Google Gemini) so your server only handles orchestration, memory, and tools.
• Telegram integration is the killer feature for OFWs — once the gateway is running, you can chat with your AI agent from anywhere, with full memory and skill persistence across devices.
• Security is built in — container hardening, namespace isolation, and dropped capabilities mean your Hermes Agent VPS runs in a sandbox that protects your system even with broad tool access.

The Problem: Why This Matters

Most AI assistants have a fundamental limitation: they only work when you are actively using them. ChatGPT closes when you close the tab. Claude forgets when you start a new conversation. Even Hermes Agent on your desktop goes offline when your computer sleeps. For Filipino professionals and OFWs who need an AI assistant that is always available — to monitor tasks, answer questions from a phone, or run automated workflows — a desktop-only deployment is not enough.

A Hermes Agent VPS deployment solves this. By running Hermes Agent on a cloud server, your AI assistant stays online 24/7, maintains persistent memory across all your devices, and is reachable through messaging platforms you already use. The agent continues learning, creating skills, and running background tasks even while you sleep, work, or travel.

The barrier has traditionally been cost and complexity. In 2026, both have collapsed. A Hostinger VPS starts at $6.49 per month, and the Hermes Agent installer handles all dependencies automatically. If you can copy and paste a command into a terminal, you can deploy a Hermes Agent VPS.

Step 1: Choose Your VPS Plan

The first decision is which VPS plan to choose. Hermes Agent needs minimal resources because the AI model runs in the cloud — your VPS only handles the agent runtime, memory storage, tool execution, and messaging gateway. Here is how the Hostinger VPS plans compare for a Hermes Agent VPS deployment:

For most Filipino professionals, the KVM 1 plan at $6.49 per month is sufficient. It gives you 4 GB of RAM and 50 GB of NVMe storage, which is more than enough for the Hermes Agent runtime, memory store, and messaging gateway. If you plan to run multiple profiles or use browser automation tools heavily, the KVM 2 at $8.79 per month provides comfortable headroom.

When selecting a server location, choose the data center closest to you. Hostinger has data centers across North America, Europe, and Asia. For users in the Philippines, the Singapore or Asian data centers will give the lowest latency. For OFWs in the Middle East, European data centers may offer better connectivity.

You can get started with Hostinger VPS hosting here. (Note: this is an affiliate link — if you sign up through it, we may earn a commission at no extra cost to you. We recommend Hostinger because it offers reliable KVM-based VPS hosting at prices accessible to Filipino professionals.)

Step 2: Set Up Your Ubuntu Server

Once you have your VPS, the next step is to set up the operating system. Hostinger offers one-click OS deployment, making this step straightforward.

2.1 Choose Ubuntu 22.04 or 24.04 LTS

From your Hostinger dashboard, navigate to your VPS and select the operating system. Choose Ubuntu 22.04 LTS or 24.04 LTS — these are the most stable and well-supported distributions for Hermes Agent. The LTS (Long Term Support) designation means you get security updates for five years.

2.2 Connect via SSH

Once the OS is deployed, connect to your server using SSH. On macOS or Linux, open your terminal:

ssh root@your_server_ip

On Windows, use PowerShell, Windows Terminal, or an SSH client like PuTTY. Hostinger provides your server IP and root password in the VPS dashboard. You can also set up SSH keys for more secure access.

2.3 Update the System

Before installing anything, update your system packages:

apt update && apt upgrade -y

This ensures you have the latest security patches and package versions.

2.4 Install Prerequisites

Hermes Agent requires Git, curl, and xz-utils on Linux. Install them:

apt install git curl xz-utils build-essential -y

The Hermes Agent installer handles everything else — Python, Node.js, ripgrep, and ffmpeg are installed automatically. You do not need to install any of these manually.

Step 3: Install Hermes Agent on Your VPS

Now for the main event. Installing Hermes Agent on your VPS is a single command:

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

The installer will download and set up everything: the Hermes Agent repository, Python virtual environment, Node.js runtime, all dependencies, and the global hermes command. This typically takes two to three minutes on a fresh VPS.

Once installation completes, reload your shell and verify your Hermes Agent VPS is ready:

source ~/.bashrc

Then verify the installation:

hermes doctor

The hermes doctor command checks your Hermes Agent VPS installation, dependencies, and configuration. If everything is green, your Hermes Agent VPS is ready to proceed. If there are issues, hermes doctor --fix will automatically repair common Hermes Agent VPS problems.

For a headless VPS deployment where you do not need browser automation, you can skip the Chromium installation:

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash -s -- --skip-browser

This saves disk space and installation time. You can always add browser support later if needed.

Step 4: Configure Your AI Model

Hermes Agent is model-agnostic — it can connect to any LLM provider. For a VPS deployment, you will use a cloud model since your server does not have a GPU. Here are the recommended options, ranked by cost and quality:

Option A: Nous Portal (Easiest — Recommended for Beginners)

hermes setup --portal

This single command logs you into Nous Portal, selects a capable model, and enables the Tool Gateway (web search, image generation, TTS, and cloud browser) in one step. The free tier is enough to get started, and a paid subscription gives access to 300+ models. This is the fastest path to a working agent.

Option B: DeepSeek V4 Flash (Best Value)

DeepSeek offers excellent tool-calling at roughly $0.14 per million input tokens and $0.28 per million output tokens. A few dollars goes a long way. Configure it with:

hermes setup

Then select DeepSeek from the provider list and enter your API key. Get your API key from the DeepSeek platform.

Option C: Google Gemini Free Tier (Free)

Google Gemini 2.5 Flash offers a free tier with good tool-calling capabilities. This is the best zero-cost option. Configure it through hermes setup and select Google as your provider. Rate limits apply on the free tier, but for personal use it is usually sufficient.

Whichever option you choose, verify the model is working:

hermes chat -q "What can you do?"

If you get a response, your Hermes Agent VPS is live and connected to an AI model. This is the moment your Hermes Agent VPS deployment becomes real — you now have an AI agent running on your own server.

Step 5: Enable Essential Tools

Tools are what make Hermes an agent rather than a chatbot. On a fresh VPS installation, enable these four tools first:

hermes tools enable web # Web search and content extraction
hermes tools enable file # Read/write files on your system
hermes tools enable terminal # Run shell commands
hermes tools enable memory # Remember preferences across sessions

After enabling tools, start a fresh session:

hermes chat

Then type /tools to verify all four tools are active. If tools do not appear, run hermes tools to open the interactive picker — tools are scoped per platform, so enable them for the CLI and Gateway separately.

Why these four tools matter on a VPS:

• Web — lets your agent search the internet and extract page content. Without this, most research tasks fail silently.
• File — read, write, and search files. Essential for any productivity task on your server.
• Terminal — run shell commands. This is what makes Hermes an agent — it can install software, manage files, and automate system tasks on your VPS.
• Memory — persistent cross-session memory. Without it, your agent forgets everything between sessions. On a VPS, this is especially important because your agent runs continuously.

Step 6: Set Up the Messaging Gateway

This is where your Hermes Agent VPS becomes a 24/7 assistant. The messaging gateway connects your agent to Telegram, Discord, WhatsApp, and other platforms, so you can chat with it from your phone.

6.1 Configure the Gateway

hermes gateway setup

This launches the interactive gateway setup wizard. Select the platforms you want to connect. For most Filipino users, Telegram is the best starting point — it is free, widely used, and has excellent bot support.

6.2 Create a Telegram Bot

To connect Telegram, you need a bot token from BotFather:

1. Open Telegram and search for @BotFather
2. Send /newbot and follow the prompts
3. BotFather will give you an API token — save this
4. Enter the token when the Hermes gateway setup prompts you

6.3 Install the Gateway as a System Service

To make the gateway survive reboots and stay running 24/7, install it as a systemd service:

hermes gateway install
hermes gateway start

Then enable linger to prevent the service from dying when you log out of SSH:

loginctl enable-linger $USER

Verify the gateway is running:

hermes gateway status

If the status shows “active” and “running,” your Hermes Agent VPS is now live on Telegram. Send a message to your bot and your Hermes Agent VPS should respond. This is the payoff — your Hermes Agent VPS is now a 24/7 AI assistant accessible from anywhere.

Step 7: Secure Your VPS

Running an AI agent on a public server requires basic security hygiene. These steps take 10 minutes and protect your server from the most common attacks.

7.1 Enable the Firewall

ufw allow ssh
ufw allow 443
ufw enable

This allows SSH and HTTPS traffic while blocking everything else. If you set up a web interface for Hermes, port 443 is already open.

7.2 Disable Root Login (Recommended)

Create a non-root user and switch to SSH key authentication:

adduser hermes
usermod -aG sudo hermes
rsync --archive --chown=hermes:hermes ~/.ssh /home/hermes/.ssh

Then edit /etc/ssh/sshd_config and set PermitRootLogin no. Restart SSH: systemctl restart sshd.

7.3 Set Up Automatic Updates

apt install unattended-upgrades -y
dpkg-reconfigure -plow unattended-upgrades

This ensures your server receives security patches automatically. For a deeper dive into server security, see our Multi-Factor Authentication Setup Guide.

What This Means for Filipino Professionals

A Hermes Agent VPS deployment is particularly valuable for three groups of Filipino professionals:

OFWs can deploy a Hermes Agent on a VPS and access it from Telegram while at work in Saudi Arabia, UAE, Singapore, or anywhere else. The agent maintains persistent memory across all interactions — it remembers your business details, family schedules, financial goals, and project contexts. You can ask it to draft emails, research topics, manage tasks, or run automated workflows from your phone, and it continues working even when you are offline. The $6.49 monthly cost is less than most OFWs spend on mobile data add-ons.

Filipino developers and entrepreneurs can use a Hermes Agent VPS as a development assistant that runs 24/7. The agent can monitor GitHub repositories, run CI/CD tasks, manage deployments, and handle server administration through the terminal tool. With profiles, you can separate work and personal contexts completely.

Small business owners can deploy Hermes Agent as a cost-effective AI assistant that automates repetitive tasks — content research, email drafting, data analysis, social media monitoring. At $6.49 per month, it is dramatically cheaper than hiring a virtual assistant or subscribing to multiple SaaS tools. For more on how AI can transform your business, read our AI Business Philippines 2026 Guide.

For a comprehensive overview of what Hermes Agent can do, including its self-improving skills system and multi-platform reach, see our Complete Hermes Agent Guide. To understand where Hermes fits among other AI agent platforms, check our Best AI Agents 2026 comparison.

Common Mistakes to Avoid

1. Running the gateway without systemd. If you start the gateway with hermes gateway start but do not run hermes gateway install, the gateway will die when you close your SSH session. Always install it as a system service and enable linger with loginctl enable-linger $USER.

2. Using an 8B model for tool calling. The Hermes community has consistently found that 8B-class models (Llama 3 8B, Gemma 3 8B) fail at reliable tool use. Hermes needs at least a 12B model for consistent tool-calling. On a VPS without a GPU, use cloud models like DeepSeek V4 Flash or Gemini 2.5 Flash instead of trying to run small local models.

3. Forgetting to enable tools. Tools are disabled by default on some platforms. After installation, run hermes tools to open the picker and enable web, file, terminal, and memory. Then run /reset in your chat session — tools are loaded at session start, so changes require a fresh session.

4. Mixing installation methods. If you install Hermes via the curl installer and then try to install it again via Docker or git clone, you will end up with two installations fighting each other. Pick one method, and if you need to switch, run hermes uninstall and remove ~/.hermes/ before reinstalling.

5. Leaving the firewall disabled. A VPS on the public internet without a firewall will be scanned and probed within minutes. Enable UFW before installing anything else, and only open the ports you actually need.

Tools and Resources

• Hostinger VPS Hosting — KVM-based VPS starting at $6.49/month. AMD EPYC processors, NVMe storage, data centers in Asia. Get started here.
• Hermes Agent Official Site — Download, documentation, and Nous Portal access. hermes-agent.nousresearch.com
• Hermes Agent GitHub — Source code, issues, and community contributions. github.com/nousresearch/hermes-agent
• DeepSeek Platform — Get API keys for the best-value cloud model. platform.deepseek.com
• Hermes Agent Discord — Community support and troubleshooting. discord.gg/nous

Summary and Next Steps

You now have a Hermes Agent running on a VPS, connected to a cloud AI model, with tools enabled and a Telegram gateway that keeps it online 24/7. The total cost is $6.49 to $8.79 per month for the VPS plus whatever you spend on AI model API access (which can be zero if you use free tiers).

Here is what to do in your first week:

1. Day 1: Send a message to your Telegram bot and test basic conversation. Ask it “What tools do you have?” to confirm everything is working.
2. Day 2: Ask your agent to do a real task — search the web for a topic and summarize the results. This tests the web and terminal tools.
3. Day 3: Create a profile: hermes profile create work. This separates your work context from personal use.
4. Day 4: Set up a second messaging platform — try Discord or WhatsApp to expand your access points.
5. Day 5: Explore skills. Ask your agent to teach you about the /learn command and start building your skill library.

Frequently Asked Questions

How much does it cost to run Hermes Agent on a VPS?
The VPS hosting costs $6.49 to $8.79 per month on Hostinger. The Hermes Agent software itself is free and open-source (MIT license). AI model access ranges from $0 (free tiers on Nous Portal or Google Gemini) to a few dollars per month (DeepSeek V4 Flash at $0.14 per million input tokens). Total cost: $6.49 to $15 per month depending on your usage.

Do I need a GPU on my VPS?
No. The Hermes Agent VPS approach connects to cloud AI models, so your server only handles orchestration, memory, and tools. A basic KVM 1 plan with 1 vCPU and 4 GB RAM is sufficient. You only need a GPU if you want to run local AI models, which is a separate and more expensive deployment path.

Can I access my Hermes Agent from my phone?
Yes. Once you set up the messaging gateway with hermes gateway setup and hermes gateway install, your agent is reachable through Telegram, Discord, WhatsApp, and other messaging platforms. You get full agent capabilities — including tool use, memory recall, and skill execution — directly from your phone’s messaging app.

Is my data safe on a VPS?
Hermes Agent stores all data (memory, skills, session history) on your own server, not on a third-party cloud. The agent runs with container hardening, namespace isolation, and dropped capabilities. By following the security steps in this guide — enabling the firewall, disabling root login, and setting up automatic updates — your VPS deployment is as secure as any commercial cloud service.

What happens if my VPS restarts?
If you installed the gateway as a systemd service with hermes gateway install and enabled linger with loginctl enable-linger $USER, the gateway will automatically restart when the VPS reboots. Your memory and skills persist because they are stored on disk. You may need to start a new chat session, but all context is preserved.

Can I run multiple Hermes Agent profiles on one VPS?
Yes. Use hermes profile create to create separate profiles for work, personal, and experimental use. Each profile has its own memory, skills, model configuration, and session history. This is useful for OFWs who want to separate their professional tasks from personal projects on the same server.