OpenClaw vs Hermes Agent: which to run
OpenClaw and Hermes Agent are both free to self-host on your own computer, both plug into whatever model you already pay for, and both reach you through Telegram, Discord, Slack and a dozen other chat apps. The real difference is who is behind them and what they charge for. OpenClaw is run by a nonprofit foundation with no paid tier at all. Hermes is run by a venture-backed company, Nous Research, whose own docs point paying users toward its hosted Nous Portal for extra models and tools. Pick OpenClaw if you want nobody's business model attached to your agent; pick Hermes if you want one subscription that reaches 300-plus hosted models without juggling API keys.
If you would rather not host anything, Mover OS is a third kind of option: $49 once, it runs inside the coding agent you already use, such as Claude Code or Codex, keeps plain files on your own machine and plans your day. It is not an always-on agent, because it stops when that agent closes.
Last verified 29 September 2026.
OpenClaw vs Hermes, side by side
| Fact | OpenClaw | Hermes Agent |
|---|---|---|
| Maker | The OpenClaw Foundation, an independent 501(c)(3), donation-funded; original creator Peter Steinberger source | Nous Research (venture-funded; TechCrunch reported a $75M round at a $1.5B valuation, July 2026) source |
| Launch date | Repository created 24 November 2025; continuous releases since (e.g. 2026.9.5 on 19 September 2026) source | Repository created 22 July 2025; public launch "since February 2026" per its own star-growth framing (64K+ stars since Feb 2026 launch, 247,436 stars by 20 September 2026) source |
| Price and plans | Free, open source, self-hosted. Official FAQ: "no paid tier, no hosted service, no token." A separate domain, getopenclaw.ai, advertises an unaffiliated-looking "$49/mo Cloud" offer that contradicts the official Foundation FAQ; not used here as an official price, flagged as a likely unofficial/SEO site rather than the Foundation's own product. source | Free and open source (MIT license) to self-host with your own model API keys. A Nous Portal subscription (for 300+ bundled models plus a Tool Gateway) is reported at Free/$20/$100/$200 per month tiers, triangulated from third-party SEO pages and TechCrunch's reported $20-$200/mo range; the official pricing page itself returned HTTP 429 to automated checks in this and the prior pass, so exact tier names and credit amounts are UNVERIFIED against the primary source. Separate Business and Enterprise tiers exist (contact sales). source |
| Availability | Global; self-hosted on hardware you control, any OS. source | Global; self-hosted, open source, no account required for the core agent (an anonymous/free-tier auth path exists for those who do want hosted models). source |
| Platforms | macOS, Linux, Windows (native and WSL2), Raspberry Pi; reachable via Discord, iMessage, Signal, Slack, Telegram, WhatsApp, Matrix, IRC, Nostr and 20+ more channels, plus mobile voice and hosted widgets. source | macOS, Linux, Windows; a desktop app; 22+ first-party messaging-platform plugins (Telegram, Discord, Slack, WhatsApp, Matrix, LINE, Feishu, Teams and more) plus generic webhook/API-server ingress. source |
| Underlying model | Model-agnostic: Anthropic, OpenAI, MiniMax, OpenRouter and others via plugin, with per-agent routing and failover; a local-only option runs entirely on-device models (llama.cpp, LM Studio, Ollama, MLX, vLLM, SGLang). source | Model-agnostic via the Nous Portal: 300+ models including Anthropic Claude, OpenAI GPT, Google Gemini, DeepSeek, Qwen, Kimi, GLM, MiniMax, Grok, and others; or local models via Ollama/llama.cpp, subject to a minimum-context-length floor that can reject small local models by default. source |
| Where memory lives | Local, on the user's own hardware, alongside session state and credentials. "State, memory, and credentials live on your hardware." A competitor's own technical comparison (OpenClaw's docs) notes memory entries carry no source-session lineage or deletion tombstones by default. source | Two local files, MEMORY.md (2,200-character budget) and USER.md (1,375-character budget), on the user's own machine under ~/.hermes/. Injected as a frozen snapshot at session start; the tool has no read action, and by its own design has no source-session lineage or deletion tombstone on entries (per a competitor's technical comparison). source |
| Always on | Depends on the user's own setup: yes if the host machine (e.g. a Raspberry Pi gateway) is left running continuously, no built-in cloud always-on layer of its own. source | Not by default; a full cron/scheduler subsystem exists for unattended, scheduled jobs when self-hosted continuously, including a script-only wake-gate mode with no LLM call. source |
| Own computer / browser | No. Runs ON the user's own hardware rather than providing a separate cloud computer; sandboxing (isolated code execution) is off by default and is a configuration option, not a given. source | No. Runs on the user's own hardware (self-hosted, like OpenClaw), with an optional cloud terminal sandbox (Modal) available through the Tool Gateway rather than a dedicated always-on cloud computer of its own. source |
| Integrations | MCP client and server, the Linux Foundation A2A 1.0 agent-to-agent protocol, Agent Client Protocol (ACP) for hosting external harnesses (including Claude Code and Codex as plugins), the AgentSkills spec, an OpenAI-compatible /v1/chat/completions and /v1/responses API, OpenTelemetry and Prometheus export. source | MCP; the Nous Portal Tool Gateway (Firecrawl web search/extract, FAL image generation, OpenAI TTS, Browser Use cloud browser automation, Modal cloud sandbox); autonomous skill authoring (skill_manage) and a background skill-consolidation "Curator". source |
| Trains on your data by default | Not applicable in the way a hosted product has a default: it is self-hosted software with no central account. "By default OpenClaw itself phones home for nothing but a daily version check; anonymous feature statistics are opt-in." source | No. Hermes collects no telemetry, usage data or analytics; conversations, memory and skills stay in ~/.hermes/ and model calls go only to the provider you configure. source |
Each of the 22 cells in this table links to its source. A row is left out when no source answers it for every product shown.
Who should pick which?
Pick OpenClaw if…
Pick OpenClaw if the idea of a venture-backed company owning your agent's roadmap bothers you. It is stewarded by an independent 501(c)(3) foundation that, by its own FAQ, sells nothing: it lists no paid tier, hosted service or token. It also runs on a Raspberry Pi for $35 to $80 one time, with no monthly software fee, if you want an always-on gateway that costs less than a month of most subscriptions.
Pick Hermes if…
Pick Hermes if you want to try dozens of frontier models (Claude, GPT, Gemini, DeepSeek, Grok and more) through one subscription instead of five separate API keys, and you're comfortable that the company behind it took a $75M round at a reported $1.5B valuation and plans to make money from paid tiers. Hermes also has a feature OpenClaw doesn't: it can write and file its own new skills automatically as you use it, reviewed by a background pass every 10 tool calls or so.
Skip both if…
OpenClaw and Hermes Agent are both not for anyone who wants a hosted agent that works with no setup: each one asks you to install, update and secure software on your own machine. If that is a dealbreaker, the alternatives pages list hosted options with their own limits.
What do users say?
“I've been using open claw for the past 3 months and it has some quirks but it works pretty well... The CLI install flow is one of the best I've seen.”
onel, Hacker News, OpenClaw
“About three weeks ago I was seduced by people singing praise to OpenClaw... It turned out to be probably the crappiest, glitchiest piece of software I've used in the past few years. Its basic onboarding workflow was completely broken, GUI was a hallucinated mess.”
nlitened, Hacker News, OpenClaw
“A friend of mine automated the lead generation and marketing function of his tiny startup using OpenClaw and a set of skills he wrote for it... They actually closed a few deals following up on leads generated by their bots.”
noufalibrahim, Hacker News, OpenClaw
“I tried OpenClaw, but holy smokes, that project is bigger than Postgres. All I want is a Pi session I can put on a server and load up with skills, that I can chat with over matrix.”
pkulak, Hacker News, OpenClaw
“I've been looking for a new cooker. I have a Hermes agent running on VM that does this. I told it to keep an eye on prices. It created a cron job that tri[ggers]...”
cassianoleal, Hacker News, Hermes Agent
“IMO this comes down to your harness. Any frontier model from a huge shop has an inherent benefit in the system you're using it in... I recommend trying Hermes Agent and setting it up fully, that's the closest you'll get to a more complete experience.”
wronglebowski, Hacker News, Hermes Agent
“I also run Hermes Agent and have set up there one agent to be responsible for the KB stuff and that one I connected to a Matrix channel and can use it from the phone and any device where I have a matrix client like Element set up.”
jeena, Hacker News, Hermes Agent
“The local model forgot things without saying so. Every model can hold only a limited amount of text at once, its context window, and in long conversations the local model was silently dropping the beginning.”
Marc Bara, Medium, Hermes Agent
Common questions
Is Hermes Agent actually free?
The core agent is free and open source (MIT license) and you can run it with your own API keys at no extra cost beyond what your model provider charges. The Nous Portal, which bundles 300-plus models and extra tools into one subscription, is reported at $20 to $200 a month, but the Nous Portal's own pricing page returned an HTTP 429 bot-detection error on every attempt to read it directly in this research, so those exact tiers are triangulated from third-party sources, not confirmed on the official page.
Does OpenClaw run in the cloud, or only on my computer?
By default it runs on hardware you control: a Mac, a Linux box, a VPS, or something as small as a Raspberry Pi. It is not a hosted cloud service the way Meta Muse or OpenAI's dots are. Sandboxing (isolating what the agent's code execution can touch) is off by default in OpenClaw and is something you turn on, not a given out of the box.
Can either of these use Claude or GPT models I already pay for?
Yes to both, in different ways. OpenClaw treats Claude, OpenAI, MiniMax and OpenRouter models as swappable plugins with per-agent routing and failover. Hermes reaches over 300 models including Claude and GPT through its own Nous Portal subscription rather than your existing Claude or ChatGPT account directly.
Which one is easier to set up?
Real user reports on both are mixed and contradictory. On Hacker News, one long-time OpenClaw user called the CLI install flow "one of the best I've seen," while another called the same onboarding "completely broken" and the GUI "a hallucinated mess." Treat both projects' setup experience as something to try yourself rather than take either camp's word for.
Neither one fits? Each agent has its own alternatives page, with every fact linked to its source. Or Mover OS, $49 once. It is the wrong choice if you want something that keeps running while your computer is off.