Best Hermes Agent VPS In 2026: The Ranked List, The Size You Need, And The Models Worth Running On It

The best Hermes Agent VPS for most people is the Hostinger KVM 2, which has 2 vCPU and 8GB of RAM for $8.99 a month on promo and comes with a one-click Hermes Agent template.

Hermes Cloud from Nous Research is my second pick if you never want to manage a server, and Oracle Cloud's Always Free tier is the best free option.

I checked every price on the provider's own page on 10 October 2026.

I run Hermes daily on my Mac and I've deployed Hermes Cloud, but I haven't run Hermes on every VPS below, so the VPS ranking is based on published specs, published prices and the official Hermes docs.

Below you'll find the ranked list, the server size you actually need, the setup steps for an Oracle VPS, and a late section on which models are worth running on that server.

The best Hermes Agent VPS options at a glance

How I ranked the best Hermes Agent VPS options

I used four tests, and you can repeat each one yourself.

The first test is whether the plan has enough memory for Hermes and its browser tools.

The second test is the real monthly price today, including the renewal price where the provider shows one.

The third test is how many steps sit between paying and messaging your agent.

The fourth test is risk, such as a long lock-in or a free server that can be taken back.

These are my own picks, not benchmark results, and the only measured numbers in this post are the model scores further down.

What Hermes Agent is and why it needs a home

Hermes Agent is the free, open-source AI agent from Nous Research, released under the MIT licence.

It runs on Linux, macOS, WSL2, native Windows and Android through Termux.

It has a messaging gateway for apps like Telegram, Discord and Slack, and it can run scheduled jobs.

Those two features only pay off when the machine is awake, and a laptop isn't always awake.

That's why people look for a VPS, which is simply a rented server that stays on.

The 7 best Hermes Agent VPS options, ranked

1. Hostinger KVM 2

The KVM 2 plan has 2 vCPU, 8GB of RAM, 100GB of NVMe storage and 8TB of bandwidth.

The promo price of $8.99 a month is paid upfront, and the page listed a renewal price of $14.99 a month.

Hostinger's own Hermes Agent cost guide says its template installs Hermes Agent in one click.

That guide recommends at least 2 vCPU and 8GB of RAM for a stable Docker deployment, which is exactly this plan.

The smaller KVM 1 has 1 vCPU and 4GB of RAM at $6.49 a month on promo, renewing at $11.99.

2. Hermes Cloud

Hermes Cloud is the hosted version of Hermes Agent, and Nous Research runs it.

The Standard size has 2GB of RAM, 2 vCPUs and 10 concurrent sessions for $0.51 a day while running.

The Pro size has 4GB of RAM, 4 vCPUs and 20 concurrent sessions for $0.99 a day while running.

A stopped agent costs $0.03 a day for storage, and inference and tool usage are billed on top.

You need a $2 minimum credit balance or an active Nous subscription to deploy one.

Thirty days of Standard running non-stop works out at about $15.30 before model usage.

3. Oracle Cloud Always Free

Oracle's Always Free tier includes Arm-based Ampere A1 compute.

Oracle's docs describe the allowance as equivalent to 2 OCPUs and 12GB of memory for Always Free tenancies.

The tier also includes 200GB of block storage and 10TB of outbound data a month.

Hermes lists Linux on aarch64 as a Tier 1 platform, so the Arm shape is supported.

The catch is that Oracle can reclaim Always Free instances it judges to be idle over a 7-day period.

4. Hetzner Cloud

Hetzner is a well-known European provider with data centres in Germany and Finland, plus US and Singapore locations.

Its regular performance range lists the CPX12 with 1 vCPU and 2GB of RAM, and the CPX22 with 2 vCPU, 4GB of RAM and 80GB of NVMe.

The prices didn't load for me on 10 October 2026, so check current pricing on Hetzner's own page.

5. DigitalOcean

DigitalOcean's Basic Droplets are the simplest to understand if you've never rented a server.

The 2GB Droplet with 1 vCPU costs $12 a month, and the 4GB Droplet with 2 vCPUs costs $24 a month.

You pay more for each gigabyte of memory than you do elsewhere on this list.

6. Contabo

Contabo's Cloud VPS 4 has 4 vCPU, 8GB of RAM and 100GB of SSD from €4.40 a month.

Contabo states that the displayed price is the effective monthly rate on a 24-month subscription.

It's the most memory for the money here, as long as you're happy to commit.

7. Vultr

Vultr is a well-known VPS provider, but its pricing page blocked my check on 10 October 2026.

I won't quote specs or prices I haven't seen, so check current pricing on Vultr's site.

What size server does Hermes Agent actually need?

The size depends on where the model runs, because Hermes is the harness and the model is the brain.

The official installation docs don't publish a minimum RAM or CPU figure.

The Hermes README says you can run it on a $5 VPS, and that holds for a light agent.

The installer adds a browser tool with a pinned Chromium by default, which is why I'd avoid 1GB plans.

Nous sizes its own smallest Hermes Cloud instance at 2GB of RAM and 2 vCPUs, so I treat 2GB as the practical floor.

How to set up Hermes Agent on an Oracle VPS

One of my viewers asked me how to do this, so here are the real steps.

First, create a compute instance in the Oracle Cloud console with a Canonical Ubuntu image and the VM.Standard.A1.Flex shape set to 2 OCPUs and 12GB of memory.

Save the SSH key the console offers, then connect from your own computer.

ssh -i ~/.ssh/your-oracle-key.key ubuntu@YOUR_SERVER_IP

Second, patch the server and create a separate non-root user to own the agent.

sudo apt update && sudo apt upgrade -y

sudo adduser --disabled-password --gecos "" hermes

sudo -iu hermes

Third, run the official install script as that user.

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

source ~/.bashrc

Fourth, run the setup wizard to choose your model, and run the doctor command to check for problems.

hermes setup

hermes doctor

Fifth, connect Telegram by creating a bot with @BotFather and running the gateway wizard.

hermes gateway setup

Set TELEGRAM_ALLOWED_USERS to your own numeric user ID, which @userinfobot will give you.

Sixth, install the gateway as a service and start it.

hermes gateway install

hermes gateway start

Seventh, switch back to the ubuntu user and enable lingering so the service survives a logout.

sudo loginctl enable-linger hermes

The Hermes docs also describe sudo hermes gateway install --system for headless hosts that should come back at boot.

Telegram uses polling by default, so the server only makes outbound requests and you don't open any inbound port.

Security basics for a Hermes Agent VPS

An agent with terminal access on a public server deserves real care.

You should run Hermes as a non-root user with no sudo rights.

You should log in with SSH keys only and leave password logins switched off.

You should allow SSH through the firewall and nothing else.

The gateway denies unknown users by default, and you should never set GATEWAY_ALLOW_ALL_USERS=true.

Your API keys belong in ~/.hermes/.env on the server, and they should never be pasted into a chat.

You can run hermes update for fixes and hermes security to audit plugins and MCP servers.

If you want private access to the dashboard, Tailscale puts the server on your own private network without opening a port.

My Agent OS on your phone guide walks through that Tailscale setup.

VPS versus local Mac versus Hermes Cloud

I build on a Mac, because that's where local models and the desktop tools live.

A server earns its place when a job has to run while the Mac is shut.

My sister post on the best Hermes Agent setup compares the full stacks.

What it honestly costs to run Hermes on a VPS

The server is usually the small part of the bill, and the model is the part that moves.

Nous Portal's free plan is $0 and includes free models only.

Its Plus plan is $20 a month and includes $22 of monthly credits.

An honest monthly budget is the server price plus whatever your model usage comes to.

On Oracle with a free model, the total can be zero, but free models get rate limited and the free server can be reclaimed.

On Hostinger, the total is $8.99 on promo for the server plus your model spend.

That's why the model you choose matters more to your bill than the VPS you choose.

Which models are worth running on a Hermes Agent VPS?

This is the part of the decision that GoldieBench can actually help with.

Hermes Agent doesn't ship its own model, so a VPS only gives it a place to run.

The brain is whichever model you connect, and on a normal VPS that should be an API model.

GoldieBench is my leaderboard, and it gives every model the same one-shot prompt to build a game, a page, a simulation or a visual, then scores the result from 0 to 10.

It doesn't measure server performance, uptime or how well a model handles a Telegram chat, so I won't pretend it does.

What it does show is how well a model follows a detailed brief and ships working output first time, which is a fair signal for agent work.

Hermes MoA is Hermes Agent's own Mixture of Agents mode, and its 8.17 shows how far the harness can go with strong models behind it.

MiniMax M3 is the one I'd look at first for an always-on agent, because it sits close to the top at a small price per token.

Claude Opus 5 has the highest single-model average in that table, and it's the brain I'd pay for on hard builds.

What about local models on the server?

Two local models on the board show why I keep local models off a normal VPS.

Both of those scores came from a Mac with Apple silicon, not from a rented CPU server.

A normal VPS has no GPU, so I'd expect those same models to run far too slowly there to be useful agents.

I haven't benchmarked any model on a VPS, so I can't give you a tokens-per-second number for one.

If you want a local brain, run it on your own machine and see the local models board and my best local model for Hermes Agent breakdown.

For API brains, my best LLMs for Hermes Agent post and the sister ranking of the best Hermes Agent model go deeper.

The free models that aren't on the board

The free models on Nous Portal change over time, and several of them have no GoldieBench score.

I won't give a model a number it hasn't earned, so test any free model on your own jobs before you rely on it.

My Solar Mini 4 and Hermes post covers one free brain I tried.

Also On Our Network

• Agent Operatorsthe full Oracle VPS setup walkthrough for Hermes Agent
• AI Income Deskwhat a Hermes server really costs a business
• AI Tool Verdictseven Hermes Agent hosts compared and scored
• agentos.guidethe Hermes VPS quick-start and how it fits an Agent OS
• agentos.guideMy Hermes Cloud guide on agentos.guide

Pick the host that matches your patience for server work, connect a model that scores well on the board, and you'll have the best Hermes Agent VPS for your own setup.