Blog / VPS sizing

Hermes Agent VPS Requirements: CPU, RAM & Disk

Hermes Agent has no single universal VPS specification. A private Telegram assistant can start small, while browser automation, scheduled jobs, multiple profiles, and team traffic need more memory and recovery headroom. Size from the workload you will actually run, keep /opt/data on persistent storage, and prove restart and channel delivery before treating the server as production.

Updated Oct 30, 20269 min read
Deployment route
  1. 01

    Define the first workload

    Separate a single-user chat, an always-on messaging gateway, and a team assistant with browser tools or multiple profiles.

  2. 02

    Size compute and memory

    Begin with headroom for Linux, Docker, the Hermes runtime, logs, and one model or tool flow, then measure real usage.

  3. 03

    Plan persistent storage

    Keep configuration, sessions, memories, skills, cron jobs, and provider credentials under /opt/data.

  4. 04

    Test the failure path

    Restart the host, confirm the gateway recovers, and send a real message before expanding access.

Four-step path

How do you size a VPS for Hermes Agent?

Start with one workload, reserve capacity for the runtime and tools, then test persistence and recovery before adding more channels or schedules.

  1. Step 01

    Define the first workload

    Separate a single-user chat, an always-on messaging gateway, and a team assistant with browser tools or multiple profiles.

  2. Step 02

    Size compute and memory

    Begin with headroom for Linux, Docker, the Hermes runtime, logs, and one model or tool flow, then measure real usage.

  3. Step 03

    Plan persistent storage

    Keep configuration, sessions, memories, skills, cron jobs, and provider credentials under /opt/data.

  4. Step 04

    Test the failure path

    Restart the host, confirm the gateway recovers, and send a real message before expanding access.

What are the minimum VPS requirements for Hermes Agent?

The practical minimum is a supported Linux host with Docker, persistent storage, secure SSH access, and enough memory to run Hermes Agent plus the tools you enable. There is no reliable one-size RAM number because browser automation and concurrent profiles change the load.

A lightweight single-user messaging assistant may fit on a small entry-level instance, but that is a starting hypothesis rather than a guarantee. Browser sessions, package installs, local tools, and several profiles can move the deployment into a larger class quickly.

  • A current Linux distribution with a supported Docker installation.
  • Enough memory for the Hermes runtime, its model client, channel gateway, logs, and one tool process.
  • Persistent disk mapped to /opt/data rather than storage inside a disposable container layer.
  • SSH or another private administration path; avoid running setup through a constrained browser terminal.
  • A backup destination outside the VPS itself.

How much CPU and RAM should you reserve?

Reserve at least two virtual CPUs and 4 GB of RAM for a comfortable always-on Hermes Agent gateway with room for logs and one companion process; use a smaller trial only when you can measure the workload.

This is a planning baseline, not an official minimum. Hermes can run on less when the workload is narrow, and it can need more when Chromium, desktop tooling, several profiles, or frequent scheduled jobs run together.

Watch resident memory during a real tool call rather than only after an idle boot. If the host begins swapping during normal work, move up a size instead of relying on swap as long-term capacity.

  • Personal chat only: start small and measure the idle and active runtime footprint.
  • Always-on gateway: leave room for restart overlap, logs, and channel reconnection.
  • Browser or computer-use tools: budget materially more memory and disk.
  • Multiple profiles or team usage: separate resource limits and monitor per process.

How much disk does Hermes Agent need?

Start with enough persistent disk for the operating system, Docker images, Hermes state, logs, and tool downloads, then add room for backups and future upgrades. The persistent /opt/data volume matters more than the raw advertised disk number.

Do not place production state only inside the container. The official container expects user data under /opt/data, so a bind mount or named volume must survive image replacement.

  • Image and package caches can grow after repeated updates.
  • Browser and computer-use tools may install large dependencies.
  • Session, memory, and skill files grow with actual use.
  • Logs need rotation so they do not consume the remaining space.
  • Keep a separate snapshot or backup copy for recovery.

Which network and security settings should a Hermes VPS use?

Expose only the ports you need, administer the host over SSH, keep API and dashboard ports private, and give the channel bot only the permissions required for its intended conversations.

  • Use key-based SSH and restrict the source address where practical.
  • Open the Hermes API or dashboard only through a VPN, tunnel, or authenticated proxy.
  • Store provider and channel secrets in the runtime configuration, not shell history.
  • Rotate a token immediately if it appears in logs or screenshots.
  • Test the bot in a private chat or test server before enabling groups.

Should you use a VPS or managed Hermes Agent hosting?

Choose a VPS when you want root access and will own updates, security, backups, and recovery. Choose managed hosting when you want the assistant online without turning its infrastructure into a separate project.

Compare total effort, not only the hourly server price. A cheaper VPS can cost more once you include setup time, monitoring, failed updates, backups, and incident response.

Use the VPS path when infrastructure control is valuable to the workflow. Use managed hosting when the assistant's work is valuable and the server itself is not.

Which primary sources support this page?

Product behavior is checked against the current deployment flow. These external links provide the upstream project or channel documentation used for setup details.

Direct answers

Frequently asked questions

Can Hermes Agent run on a 2 GB VPS?

It may run a narrow personal workload, but 2 GB leaves little room for Docker, logs, model tools, updates, or browser processes. Measure the real workload and treat low-memory results as a benchmark, not a production guarantee.

Does Hermes Agent need a GPU?

No for the standard path that calls hosted model APIs. A GPU is relevant only when you deliberately run a compatible local model, which is a separate infrastructure and performance decision.

Which VPS operating system should I use?

Use a current Linux distribution supported by Docker and keep it patched. The exact distribution matters less than a reproducible Docker setup, persistent storage, firewall rules, and a tested recovery path.

What must be backed up on a Hermes VPS?

Back up the persistent Hermes data directory, including configuration, provider settings, sessions, memories, skills, cron jobs, and hooks. Store the backup outside the VPS and test a restore.

Key takeaways

  • Hermes Agent has no single universal VPS specification. A private Telegram assistant can start small, while browser automation, scheduled jobs, multiple profiles, and team traffic need more memory and recovery headroom. Size from the workload you will actually run, keep /opt/data on persistent storage, and prove restart and channel delivery before treating the server as production.
  • Start with one model, one channel, and one controlled test conversation.
  • Use managed hosting when deployment speed matters more than operating the server yourself.

Put Hermes Agent in the channel where the work already happens.