Hermes vs OpenClaw: Which Self-Hosted Agent Should You Run?
Hermes Agent vs OpenClaw for self-hosters: telemetry, sandboxing, apps, chat channels, skills and setup compared, with a clear verdict on which to run.
At a glance
| Bot | License | Chat apps | |||||
|---|---|---|---|---|---|---|---|
| Hermes Agent | MIT | 261.7K | Yes | No | Telegram, WhatsApp, iMessage, Discord, Slack | From $20/mo | |
| OpenClaw | MIT | 394.2K | Yes | No | Telegram, WhatsApp, iMessage, Discord, Slack | Free |
Checklist
| Hermes Agent | OpenClaw | |
|---|---|---|
| Open-source license | Yes | Yes |
| Runs local models | Yes | Yes |
| Sandboxed by default | No | No |
| No telemetry by default | Yes | No |
| Lives in your chat apps | Yes | Yes |
| Works while you're away | Yes | Yes |
Hermes Agent and OpenClaw are the two most-starred open-source personal agents we track, and they’re closer than most “vs” pairs. Both are MIT licensed, free, model-agnostic, run on your own hardware, and live in your chat apps. Neither sandboxes commands by default. The differences are in defaults, reach and philosophy.

The short version
Run OpenClaw if you want the widest reach: native apps on macOS, Windows, Linux, iOS and Android, 29 chat channels, 142 official plugins and a one-click cloud deploy.
Run Hermes Agent if you care more about what the agent does over time and what it sends home: it writes and refines its own skills, runs separate agent profiles on one server, and shares no metrics unless you opt in.
If you’re already on OpenClaw, trying Hermes costs little, because it imports OpenClaw settings, memories and skills in one command.

Maturity and momentum
OpenClaw came first, released November 24, 2025, and had 394,193 GitHub stars when we checked on November 8, 2026. It’s now stewarded by the nonprofit OpenClaw Foundation. Hermes Agent launched February 25, 2026 from Nous Research and reached 261,682 stars by the same date. OpenClaw is TypeScript on Node; Hermes is Python.
Telemetry
This is the clearest difference. OpenClaw contacts its project server daily by default, sending your version, OS, Node.js version and CPU architecture, an identifier-free install count and update reports; the server also records approximate location. Feature statistics are opt-in, and update.checkOnStart: false turns the rest off.
Hermes Agent’s shared metrics are off by default and only sent if you choose to share them when asked. Its update check compares against the GitHub repository. Nous Research only sees your conversations if you pick its Nous Portal as your model provider.
Sandboxing
A tie, and not a good one. OpenClaw’s sandbox is off by default and the agent can run host commands unless you turn on approvals. Hermes Agent’s terminal backend defaults to local, with approval modes and container isolation as options.
Whichever you pick, turn isolation on before you connect your email. Or run either one inside NemoClaw, NVIDIA’s sandbox stack, which supports both. Our Is OpenClaw safe? guide covers the risks.
Apps and install
OpenClaw installs with a script, npm, desktop apps, Docker or a one-click Railway deploy. It needs Node 24.16+ or 26.1+, which the installer provisions if missing. It has native iOS and Android apps.
Hermes Agent installs with a script, desktop apps for macOS (Apple Silicon only) and Windows, or Docker. Linux has no desktop package, and Android is a command-line install through Termux. There’s no iOS app, so on your phone you use a chat app.
Channels
Both cover Telegram, WhatsApp, Discord, Slack and iMessage. OpenClaw lists 29 channels, including Microsoft Teams. Hermes lists 20+, including Signal, email and SMS, but iMessage needs a separate BlueBubbles server on a Mac.
Skills, memory and multiple agents
OpenClaw relies on a large ecosystem: community skills plus 142 official plugins. The flip side is that malicious third-party skills are a known risk.
Hermes takes the opposite approach. After complex tasks it writes reusable skills itself and refines them as it uses them, keeps a curated memory and builds a profile of you across sessions. Its profiles run several agents, each with its own model, memory and skills, and they can talk in group chats. OpenClaw has multi-agent routing too, but self-improvement is Hermes’ headline.
Cost
Both are free, and you pay your model provider either way. Heavy OpenClaw users have reported spending over $100 a day with expensive models. Hermes offers an optional Nous Portal subscription bundling models and hosted tools like web search and image generation; without it, those tools need your own keys. Both run local models.
Verdict
- Pick OpenClaw if you want phone apps, 29 channels and 142 official plugins, a Linux desktop app, or a one-click cloud install, and you’ll switch off the update check yourself.
- Pick Hermes Agent if you run a server, want no telemetry out of the box, use Signal or email as your main channel, or want an agent that builds its own skills.
- Pick neither as-is if you won’t configure a sandbox. Look at ZeroClaw or the other OpenClaw alternatives and Hermes Agent alternatives that isolate commands by default.
Pricing
Hermes Agent
| Open source | Free |
|
|---|---|---|
| Nous Portal Plus | $20/mo |
|
| Nous Portal Super | $100/mo |
|
| Nous Portal Ultra | $200/mo |
|
OpenClaw
| Open source | Free |
|
|---|
Hermes Agent
Strengths
- Free and MIT licensed, and runs on hardware you control
- Writes and refines its own skills from experience, on top of persistent memory and cross-session search
- One gateway for 20+ chat platforms, including Telegram, Discord, Slack, WhatsApp, Signal and email
- Run several separate agents as profiles, each with its own model, memory and skills, and let them talk in group chats
- Model-agnostic, with a one-command import from OpenClaw
Weaknesses
- Setup expects comfort with the command line and with running your own server
- Model costs are on you unless you use free models, and hosted web search, images and speech need a paid Nous Portal plan or your own keys
- iMessage needs a separate BlueBubbles server on a Mac
- No native mobile app; Android support is a command-line install through Termux
OpenClaw
Strengths
- Free and MIT licensed, with no paid tier
- Runs on your own hardware, and state, memory and credentials stay there
- Works in 29 chat channels, including WhatsApp, Telegram, iMessage, Slack, Discord and Microsoft Teams
- Model-agnostic: 64 model and media providers, plus local models
- Large ecosystem of skills and 142 official plugins, and very active development
Weaknesses
- Model costs are on you and can run high: testers and users have reported spending over $100 in a day with expensive models
- Sandboxing is off by default and the agent can run commands on the host unless you turn on approvals
- Setup and upkeep expect comfort with the command line; maintainers warn it is risky for non-technical users
- Broad access to email, messages and the system makes prompt injection and malicious third-party skills a real risk
Features
| Hermes Agent | OpenClaw | |
|---|---|---|
| Web search | Yes | Yes |
| Image generation | Yes | Yes |
| Voice mode | Yes | Yes |
| File uploads | Yes | Yes |
| Memory | Yes | Yes |
| Takes actions for you | Yes | Yes |
| App integrations | Yes | Yes |
| Works on its own | Yes | Yes |
| Messages you first | Yes | Yes |
| Runs several bots | Yes | Yes |
Sources
Hermes Agent
- NousResearch/hermes-agent on GitHub (MIT, v0.23.0)
- Hermes Agent homepage and Nous Portal pricing
- Hermes Agent documentation
- Hermes Agent docs: Profiles, running multiple agents
- Hermes Agent docs: Bot Mode
- Hermes Agent docs: Voice mode
- Hermes Agent docs: Vision and image paste
- Hermes Agent docs: iMessage via BlueBubbles
- Nous Research on X: Meet Hermes Agent (February 25, 2026)
- Hermes Agent docs: Session heartbeats and cron watchdogs
- Hermes Agent docs: FAQ (Is my data sent anywhere?)
- Hermes Agent docs: Installation (desktop packages, install script, Docker)
- Hermes Agent docs: Platform support (no Linux desktop package; PyPI installs unsupported)
- Hermes Agent docs: Docker setup
- Hermes Agent docs: Configuration (terminal backend defaults to local)
- Hermes Agent docs: Security (approval modes, optional container isolation)
- Hermes Agent docs: Local models
- Hermes Agent example config (shared metrics off by default)
OpenClaw
- OpenClaw homepage and Foundation section
- OpenClaw integrations: channels, providers and plugins
- openclaw/openclaw on GitHub (MIT, v2026.11.1)
- OpenClaw docs: Sandboxing (off by default)
- Wikipedia: OpenClaw
- Notebookcheck: OpenClaw can burn through hundreds of dollars per day
- OpenClaw docs: Memory overview
- OpenClaw docs: Automations
- OpenClaw docs: Chat channels
- OpenClaw docs: iMessage
- OpenClaw docs: Heartbeat, now a default Proactive check job
- OpenClaw docs: Multi-agent routing
- OpenClaw docs: Platforms (macOS, Windows, Linux, iOS and Android apps)
- OpenClaw docs: Tools (exec, browser, web search)
- OpenClaw docs: Audio and voice notes (inbound transcription)
- OpenClaw docs: Image and media support (inbound media)
- OpenClaw docs: Install (system requirements, installer script, npm, desktop apps, Docker)
- OpenClaw docs: One-click deploy on Railway
- OpenClaw docs: Usage telemetry and update checks (on by default, feature statistics opt-in)
- OpenClaw docs: Ollama provider (local models)

