Hermes Agent is Nous Research’s MIT-licensed self-improving personal agent: a Python gateway process that lives in your chat channels, builds skills from its own experience, and runs on everything from a $5 VPS to serverless sandboxes, at about 253k stars the biggest runtime launch since OpenClaw.
Hermes’ bet is that the runtime winner is decided by the learning loop, not the channel list: an agent that curates its own memory, writes its own skills, and models you across sessions compounds while the others merely answer.
What it is #
Install with a curl script (the installer bundles uv, Python 3.11, Node.js, ripgrep, ffmpeg, and an isolated Git Bash on Windows; the official site now also ships desktop installers for macOS 12+ and Windows 10/11), and one gateway process serves Telegram, Discord, Slack, WhatsApp, Signal, Email, and a CLI, with voice-memo transcription and conversation continuity across platforms.
The learning loop is the differentiator: it creates skills autonomously after complex tasks, improves them during use, nudges itself to persist knowledge, searches its own past sessions (FTS5 with LLM summarization), and builds a user model through Honcho, compatible with the agentskills.io open standard.
Any model goes in (Nous Portal, OpenRouter, OpenAI, custom endpoints) and switching is a hermes model command with no code changes.
Execution spans seven backends (local, Docker, SSH, Singularity, Modal, Daytona, Vercel Sandbox), with serverless persistence that hibernates when idle, and a built-in cron scheduler delivers reports to any platform.
Status #
Massive and fast-moving. 253,488 stars and 54,982 forks as of 2026-10-14 since creation on 2025-07-22, MIT, and as of 2026-10-23 pushed that day with 50,630 open issues, a support surface bigger than most projects’ users.
Releases shipped as date-stamped tags through v2026.9.24 (2026-09-24), with September alone seeing five tags, and on 2026-10-08 the project cut v0.21.6, the first semver tag and the first release of a new stable release pipeline (one attempt ref, a tested Docker image, and a receipt tag at publish), rolling up about 2,100 pull requests since v0.21.5 with the curated notes deferred to v0.22.0, and v0.21.7 followed on 2026-10-12 as the line’s second tag, with v0.22.0, “The Threshold Release”, landing 2026-10-14 (about 6,080 merged pull requests since v0.21.0, a guided first-run setup chat on a free trial model) and v0.22.1 and v0.22.2 following the same day and on 2026-10-15, with v0.22.3 and v0.22.4 following on 2026-10-21 and v0.22.5 landing 2026-10-23 (about 81 pull requests since v0.22.4, curated notes again deferred to v0.23.0).
The 52-point launch thread landed June 2026, and a real tool ecosystem followed (a 28-point Grafana observability integration, a Fleet console for managing Dockerized Hermes agents, XMPP interop with OpenClaw, and a community gadget SDK building ESP32 voice and text devices, 302 stars as of 2026-10-07).
The governance record is the caution: a May 2026 thread documents Nous editing a GitHub issue to remove plagiarism claims about Hermes Agent (
issue 10232), so origin claims around this project deserve independent reading.
Strengths #
- The learning loop (skills that write and improve themselves) is the most complete self-improvement story in the category.
- Seven execution backends with serverless idle pricing solves the always-on cost problem that keeps personal agents expensive.
- Model-agnostic to the point of indifference, with a one-command switch.
- Nous Research’s model portfolio gives it a first-party provider path the -claw tools lack.
Cautions #
- Roughly 50,000 open issues is triage weather, not a queue; expect to read code when things break.
- The plagiarism-claim edit is a stewardship red flag worth weighing before making it your memory keeper.
- A self-curating memory compounds errors as efficiently as insights; the loop needs supervision, not just trust.
- Windows installs bundle a lot of machinery (uv, Node, Git Bash), which is surface area to audit.
Pricing #
The runtime is free and open source under MIT. Paid plans now exist through Nous Portal, and as of 2026-09-22 the prices are public on the official site: Free at $0, Plus at $20/month, Super at $100/month, and Ultra at $200/month, where each paid tier bundles monthly credits with a 10 percent bonus ($22, $110, and $220 respectively), 200+ models, and hosted tool use. Or bring your own provider keys, with serverless backends billed by the platform when idle-cheap.
Price history #
| Date | Plan | Change | Source |
|---|---|---|---|
| 2026-09-18 | Nous Portal | Paid plans introduced: Free $0, Plus $20/mo, Super $100/mo, Ultra $200/mo, each paid tier bundling monthly credits with a 10 percent bonus; the runtime stays MIT free. |
Compared to #
- OpenClaw: the other giant, ecosystem-first; choose OpenClaw for channels and companions, Hermes for the learning loop and model freedom.
- NanoClaw: the auditable minimalists’ pick; Hermes is the opposite trade, a lot of machinery you must trust.
- Paperclip: if Hermes is an employee, Paperclip is the company that would manage several of them.
Bottom line #
Recommended for engineers who want their assistant to get measurably better at their job and will supervise the loop. Not for minimal-footprint deployments or anyone uncomfortable with an agent that edits its own memory. The disagreeable claim I will defend: the channel-list era of this category is over and the memory-loop era decides the winner, and Hermes is a year ahead of the -claw family on exactly that axis.
Changes #
- 2026-08-27 - Created as an Assistant runtimes note after the control-planes alternatives scan.
- 2026-09-09 - Added paid tiers (Free, Plus, Super, Ultra via Nous Portal), retiring the no-paid-tier claim.
- 2026-09-12 - Paid tier prices went public on the official site, retiring the bot-wall claim.
- 2026-09-20 - Added the Price history section tracking price changes in a table, per the new owner rule.
- 2026-09-22 - Recorded the near-daily release-tag cadence (v2026.9.7 through v2026.9.21 in September), with refreshed adoption numbers and portal tiers re-checked unchanged.
- 2026-09-24 - Recorded v2026.9.24, the fifth September tag, with refreshed adoption numbers and the Portal tiers (Free, Plus $20, Super $100, Ultra $200) re-checked unchanged on the official site.
- 2026-10-07 - Added the NousResearch/hermes-agent star history chart to the Status section.
- 2026-10-07 - Added the community hermes-gadget-sdk (ESP32 voice and text devices for the agent, 302 stars) to the ecosystem record, with refreshed adoption numbers.
- 2026-10-10 - Recorded the release-scheme change: v0.21.6 (2026-10-08) is the first semver tag and the first cut of a new stable release pipeline, ending the date-stamped scheme and the early-October tag gap, with refreshed adoption numbers and the Portal tiers re-checked unchanged.
- 2026-10-11 - Recorded the official desktop installers (macOS 12+, Windows 10/11) joining the curl install path and Email joining the official channel list, with refreshed adoption numbers and the Portal tiers (Free $0, Plus $20, Super $100, Ultra $200) re-checked unchanged on the official site.
- 2026-10-13 - Recorded v0.21.7 (2026-10-12), the second tag on the new stable release line, with refreshed adoption numbers and the Portal tiers re-checked unchanged on the official site.
- 2026-10-16 - Recorded the v0.22 line: v0.22.0, “The Threshold Release” (2026-10-14, about 6,080 merged pull requests since v0.21.0), v0.22.1 (2026-10-14), and v0.22.2 (2026-10-15), with refreshed adoption numbers.
- 2026-10-21 - Rotation re-verification: every cited URL re-fetched live (repo and releases APIs, the v0.21.6 release notes, the official site, both HN threads, the plagiarism-edit issue, the gadget SDK repo), v0.22.2 (2026-10-15) confirmed still the latest release, the seven-backend list and the Portal tiers (Free $0, Plus $20, Super $100, Ultra $200) unchanged, and the issue-10232 edit record intact.
- 2026-10-23 - Recorded the v0.22 line continuing: v0.22.3 and v0.22.4 (both 2026-10-21) and v0.22.5 (2026-10-23, about 81 pull requests since v0.22.4, curated notes again deferred to v0.23.0), with refreshed adoption numbers.
See also #
- OpenClaw - the ecosystem giant it chases
- Assistant Runtimes Feature Matrix - the category compared
- File-based agent memory - the memory conventions its loop automates
- Agent Skills open standard - the standard its self-written skills follow
References #
https://github.com/NousResearch/hermes-agent - README: learning loop, backends, channels, install
https://api.github.com/repos/NousResearch/hermes-agent - stars, forks, issues as of 2026-10-10
https://hermes-agent.nousresearch.com - official site and docs, including the Hermes Free/Plus/Super/Ultra plan tiers
https://news.ycombinator.com/item?id=48419000 - the 52-point launch thread