Starting today, Hermes Agent by Nous Research is deployable directly from the OpenClaw Launch dashboard. The inline tabbed deploy card now offers a one-click choice — launch an OpenClaw instance as before, or switch the tab to Hermes and deploy a managed Hermes agent with the same single click. Both frameworks share the same Lite and Pro plans, the same dashboard, and the same BYOK plumbing.
Why Both, Not Either
OpenClaw and Hermes answer different questions. OpenClaw excels at channel operations — configure once, reach Telegram, WhatsApp, WeChat, Discord, Feishu, and web chat from one skill set. Hermes brings Nous Research''s self-improving learning loop and persistent memory into the same operator''s hands. Running both on one dashboard means users no longer have to pick a side; the managed surface, billing, and credentials infrastructure are shared.
What Ships at Launch
The Hermes deploy path on OpenClaw Launch today supports:
- Platforms: Telegram, Discord, and Slack, with per-chip connect and disconnect on the instance card. Additional platforms (Signal, Matrix, SMS, email, iMessage, Mattermost, DingTalk, WeCom, QQ, Home Assistant, Webhooks) remain on the roadmap.
- Models: A curated OpenRouter text-model picker — Gemini 2.5 Flash (default), Claude Sonnet 4.6, GPT-5.4 Mini, DeepSeek v3.2, and MiniMax M2.7 — all swappable from the card without redeploying.
- Image generation: Eight FAL.ai models (Flux 2 Klein 9B by default) via user-supplied
FAL_KEY, injected at runtime. - Search providers: Tavily, Exa, Parallel, or Firecrawl, all bring-your-own-key.
- Execution: Full-execution mode is on by default, matching OpenClaw''s out-of-box behavior. Flip to manual approvals from the toggle on the card.
- Open Chat: Each Hermes instance exposes its OpenAI-compatible API server on a second allocated port. A lightweight chat modal on the dashboard streams responses directly through a signed proxy, so users get a web chat surface without bringing their own frontend.
- Auto-sleep and wake: Idle Hermes instances sleep on the same schedule as OpenClaw. A dashboard wake button and the first inbound message both restore them.
- Open Dashboard: Hermes''s own configuration panel is reachable from a primary CTA on the card, served through the existing Caddy gateway wildcard.
Pricing
No new tier. Hermes instances count as one instance slot on the same plans OpenClaw users already know — Lite at $6 per month, Pro at $20 per month. A paid plan covers either framework interchangeably, and a free trial is available without a card.
How It Compares
Readers evaluating Hermes hosting options can start at the framework comparison to see where Hermes fits alongside OpenClaw, or the DeployHermes comparison for a direct look at managed Hermes hosting choices. The Hermes hosting page covers feature cards and pricing at a glance.
Get Started
The step-by-step walkthrough lives at the Hermes deploy guide, with prerequisites, the dashboard form, and post-deploy controls. Existing subscribers can open the OpenClaw Launch dashboard, switch the deploy card to the Hermes tab, and deploy in under a minute.