LM Studio is a free desktop application (Windows, macOS, Linux) that lets you browse, download, and run quantized LLMs locally from Hugging Face. It provides a ChatGPT-like UI, a local OpenAI-compatible API server, and supports models like Llama, Mistral, Phi, Qwen, and hundreds more — all running 100% on your hardware with no data sent to the cloud.
Free and open-source. Hardware costs only.