LM Studio

LM Studio is the most popular desktop application for running large language models locally. It turns any capable Mac, Windows, or Linux machine into a private AI workstation — no terminal knowledge required, no data ever leaving your computer.

**Discover and run models**: LM Studio includes a built-in model browser that lets you search, download, and run models from Hugging Face in seconds. Supported model families include GPT-OSS, Qwen3, Gemma 3, DeepSeek R1, Llama, Phi, Mistral, and hundreds more. GGUF and MLX (Apple Silicon) formats are both supported.

**Local API server — OpenAI-compatible**: LM Studio runs a local server with an OpenAI-compatible REST API. Any app or script that works with OpenAI's API can be pointed at localhost:1234 and work immediately — no code changes. This makes it trivial to swap in local models for development, testing, or cost reduction.

**LM Link — remote access**: The new LM Link feature lets you connect to remote LM Studio instances, treating remote GPU machines as if their models were local. Useful for teams where one machine hosts the heavy models.

**Developer tooling**: - JavaScript SDK: npm install @lmstudio/sdk - Python SDK: pip install lmstudio - MCP (Model Context Protocol) client support - CLI via the lms command

**Completely free**: LM Studio is free for personal and commercial use. The application is closed-source but the SDKs are open. It is the go-to tool for developers, researchers, and privacy-conscious users who want production-quality LLM inference without recurring API costs.