Ollama — Run large language models locally

One-command local serving of open models with an OpenAI-compatible API — private, free inference on your own hardware.