LM Studio

LM Studio wraps local LLM inference in a desktop application for macOS, Windows, and Linux. It provides a graphical interface for discovering and downloading open models from Hugging Face, chatting with them offline, and serving them to other applications through a local OpenAI-compatible API. Under the hood it runs llama.cpp for GGUF models and Apple's MLX engine on Apple Silicon, detecting hardware and suggesting settings such as GPU offload automatically. Recent releases add agentic features, on-device voice transcription, and document chat. The application itself is proprietary but free for personal use, with open source companion SDKs for Python and JavaScript and a command line tool for headless operation. Typical users range from people who want a private chat client running entirely on their own machine to developers prototyping against local models before deploying elsewhere.