Open core. Self-host it in minutes.
The full RAGSuite pipeline — crawl, upload, chat, search and embeddable widgets — is free and open under Apache 2.0. Run it on Docker, point it at local or hosted models, and build on a REST API and an open MCP server and client.
curl https://your-instance/api/v1/search \
-H "Authorization: Bearer $RAGSUITE_KEY" \
-d '{"project":"handbook","query":"liability cap?"}'
# → answer + sources[], every claim citedOne platform, three ways to ship it.
Citation-backed Search and Assistant, plus an open connector ecosystem — all in the Community Edition.
Liability is capped at the fees paid in the prior 12 months 1, and the cap doesn’t apply to gross negligence 2.
Practitioner features free. Management and compliance, paid.
The boundary is honest and public: everything a developer needs to build is in the Community Edition under Apache 2.0. Enterprise adds what organisations need to govern it at scale — nothing is artificially crippled.
Start free. Scale to Enterprise.
If you build with it, it’s free. If you govern it, that’s Enterprise. The Community Edition is a complete product under Apache 2.0 — nothing artificially crippled; Enterprise adds what an organisation needs to run it at scale.
Community
- Full pipeline: crawl, upload, chat, search, widgets
- Entire connector & MCP ecosystem
- All LLM providers, incl. local Ollama
- REST API, keys, webhooks · Docker deploy
- Unlimited users · unlimited projects
Enterprise
- SSO/SAML/OIDC · RBAC · teams & orgs
- Full audit + compliance exports
- Compare Models · deep query tracing
- Advanced analytics · mobile app (Beta)
- Enterprise services (white-label, SLA, CSM) by agreement
Everything you need to ship — for €0.
Apache 2.0, forever. Unlimited users and unlimited projects.
Full pipeline: crawl websites (depth 1–5) and upload PDF, DOCX and TXT.
AI Search and AI Assistant — citation-backed, as embeddable widgets.
The entire connector & MCP ecosystem — Gmail, an open MCP server and client, and n8n (Beta).
All LLM providers — OpenAI, Anthropic, Mistral, Gemini — and local Ollama.
REST API, keys and webhooks · Docker deploy.
Feedback collection, basic dashboard, basic audit (30 days), 2FA and system health.
Self-host with one command.
A signed install script — Docker-native, API-first. Download it, read it, run it: RAGSuite stands up on your own infrastructure in minutes, no year-long integration project. Versions are pinned (never :latest) and open core means no lock-in — your data stays yours, exportable anytime.
Every major model — or none of them.
RAGSuite is model-agnostic. Reach for a frontier API, an EU-sovereign provider, or keep everything on-premise with local models via Ollama — switch per project, any time. Your data never leaves your infrastructure, and never trains anyone else’s model.
Compare where each model runs → Point RAGSuite at any OpenAI-compatible endpoint.
An API-first platform, documented and open.
Wire RAGSuite into your own apps and automations — or extend it.
Open MCP server & client
Expose your knowledge over the Model Context Protocol, or consume MCP sources — part of the open connector ecosystem.
Read the docs →Source on GitHub
Apache 2.0 core. Star it, fork it, open an issue, or send a PR under the CLA.
View on GitHub →Apache 2.0 at the core — and staying that way.
The Community Edition core is Apache 2.0 from the first public commit. Enterprise features live in a separate, source-available /ee directory under a commercial licence. Contributions come in under a CLA, so the project stays clean and re-licensable.
Developer questions, answered
Is the Community Edition really free?
Yes — Apache 2.0 and free forever: the full pipeline, the entire connector & MCP ecosystem, every LLM provider including local Ollama, the REST API and Docker deploy, with unlimited users and unlimited projects.
What’s Enterprise, then?
The features organisations need to govern AI at scale: SSO/SAML/OIDC, RBAC with organisations, teams and users, full audit and compliance exports, Compare Models and deep query tracing, white-label and SLA support. See pricing for the full split.
Can I run it completely offline?
Yes. With local models via Ollama, RAGSuite needs no outbound internet, and Enterprise unlocks with a signed, offline licence key — no activation server.
Which models can I use?
OpenAI, Anthropic, Mistral or Gemini under your own keys, or fully local models via Ollama. Your data never trains anyone else’s model.
Where are the docs and the repo?
Documentation lives at docs.ragsuite.de and the API reference at api.ragsuite.de; the source is on GitHub. Links are wired across the site as those go live.
“RAGSuite gave our industrial clients a sovereign AI they actually own — citation-backed, self-hosted, and defensible to their auditors.”
Evaluate RAGSuite on your own infrastructure.
Self-host the Community Edition in minutes, or talk to us about Enterprise. Either way, your data never leaves the building.