LM Studio is a desktop application for running large language models locally, on the user's own computer instead of on a hosted provider's servers. Its maker is Element Labs, Inc. The current app, LM Studio Bionic, adds an AI agent that creates and edits documents, writes code and controls the computer.
Users download open models inside the app and use them for chats or agent tasks. The models run on the LM Studio runtime, which uses llama.cpp and MLX (Apple's machine learning framework for Apple silicon) underneath. For developers, LM Studio starts a local server with its own REST API and with OpenAI-compatible and Anthropic-compatible endpoints, so software written for those hosted APIs can call a local model instead. TypeScript and Python SDKs and a command-line tool (lms) download models, start the server and script workflows; llmster, the core packaged as a daemon, runs on servers without the graphical app.
Running local models is free. Paid subscriptions add US-hosted cloud inference with open models, with zero data retention: prompts and responses are not stored by the provider.