Ollama

Ollama is the simplest way to run open models like Llama, Mistral, and Gemma entirely on your own machine. When you run a model locally, every prompt is processed on-device, so nothing you type is sent to a company's servers, logged, or used for training. Ollama now also sells cloud plans, and its cloud models run on Ollama's servers, so keep to local models if privacy is the point. It exposes a clean command line and a local REST API so other apps can plug into it without ever touching the cloud. The project is MIT-licensed and openly developed and it keeps working offline once a model is downloaded. You will want a reasonably modern machine with several gigabytes of RAM and a GPU helps but that hardware buys you total control over your data.