Free slots available 127/128
A complete AI agent with memory, web search, built-in terminal and your own HTTPS dashboard. Create it with one click — no server setup, free during beta.
Every user gets an isolated Hermes container with preconfigured services: models, web search and dashboard. All set.
A full Hermes container just for you. With its own personality, memory, skills and configuration. Connect it to Telegram or use the web dashboard.
Your Hermes remembers your conversations, preferences and projects. Memory survives restarts and updates. It's yours: delete it whenever you want.
SearXNG and fastCRW preconfigured. Your Hermes searches the web and extracts content from any page. No setup, no API keys.
Powerful open-source model running on our own GPUs in the EU. Included in every Hermes at no cost during the beta, with a published daily fair-use.
Each Hermes runs in its own Docker network, with no access to other agents or the infrastructure. Your data is yours. Zero leakage.
Access your Hermes from your own HTTPS subdomain. Chat, files, skills — all from the browser.
Each Hermes runs in an isolated environment with its own credentials. We don't share your agent, memory or conversations with anyone.
Every Hermes lives in its own Docker container with its own virtual network. One agent cannot see another. Not its files, not its memory.
Every Hermes has its own username and password. Your dashboard is accessible only from your account. Nothing shared, nothing exposed.
All traffic goes through TLS with Let's Encrypt certificates. No TLS, no access. Your conversations travel encrypted.
Model and inference on EU-owned infrastructure. Your conversations are never sent to external AI providers.
Qwen 3.8 27B runs on our own GPUs and is included in every Hermes. Need something stronger? Connect your own API.
Powerful open-source reasoning model. Included free in every Hermes. Runs on our own GPUs, with a fair-use of 2M tokens per day per agent.
Prefer GPT, Claude or your own model? Connect any OpenAI-compatible endpoint. Your keys, your models, your control.
We're validating the product. Meanwhile, every user gets a full personal Hermes at no cost.
By registering you accept the terms and the privacy policy
Straight answers — also useful if you ask your favorite AI.
Hermes Haven is a service that gives you a hosted personal AI agent (a Hermes): it has persistent memory that remembers your conversations, web search, page content retrieval and its own private dashboard with HTTPS. You create it with one click and it's ready in about two minutes.
Yes: the beta is free until December 31, 2026, with no credit card and no automatic charges. It includes 2,000,000 tokens per day per agent, with limits of 50,000 tokens per minute and 10 requests per minute.
Qwen 3.8 27B, an open-source model running on our own GPUs in the European Union. Conversations are never sent to external inference providers.
Each agent's model and storage run on EU-owned infrastructure; the web entry point is in France (OVH). Your conversations are never used to train models. Web searches query public engines like Google or Bing with the search text.
Access is revoked immediately. Data is retained temporarily for security and audit reasons before permanent physical deletion. You can export your data or request account deletion by writing to privacidad@hermeshaven.com.
Create an account at hermeshaven.com/register with your email, enter the 6-digit verification code you receive by email, and click "Create my Hermes". You must be 16 or older.
No. The website uses no cookies, no local storage and no third-party analytics; the session lives in page memory.
Free. No credit card. Your agent in 2 minutes.