The stack is OSS-first by design — FastAPI, Postgres with pgvector, Redis, Celery, and Ollama for inference — and it runs on your own machine or your own server. What the paid plans buy is hosting, support and the managed integrations. They do not buy the right to use the software, and there is no feature held hostage behind a licence check.
The same applies to model costs. On the default configuration Ollama runs inference locally, so there is no per-token vendor bill at all. Point Founder OS at Anthropic, Google or an OpenAI-compatible endpoint and you supply your own key and pay that provider directly — Founder OS does not resell tokens or mark them up.
If you want to see what you would be running before you decide, the feature breakdown describes each subsystem, and the integrations pages say exactly which tools connect today and in which direction data flows.