Ollama (MIT licensed) plus Open WebUI (also open source, self-hosted) gives you a complete local AI setup that’s open end to end — model runtime, API, and the chat interface. Nothing proprietary required, nothing phones home.

Wrote a step-by-step guide for getting this running on Windows: installing Ollama, pulling models, setting up Open WebUI through Docker, GPU offloading, and the env vars that actually matter (OLLAMA_NUM_GPU, OLLAMA_KEEP_ALIVE, etc). I also cover LM Studio as an alternative if you want a desktop app instead, though that one’s closed-source — wanted to flag that distinction since it matters here.

It’s a paid guide ($15) if you want the full walkthrough: https://indigonynja.gumroad.com/l/jqwisc

Always happy to talk Ollama config in the comments regardless of whether anyone buys anything.