LM Studio

LM Studio lets you find, download, and run large language models on your own Linux, Windows or macOS, system. Find models through its built-in search, download their weights, adjust loading settings and chat in a desktop interface. LM Studio supports GGUF models through llama.cpp, plus MLX models on Apple Silicon Macs.

Developers can start LM Studio’s local model server and connect applications through its REST API, OpenAI-compatible endpoints or Anthropic-compatible endpoints. The lms command-line tool can download models, manage the daemon and control the server. Python and TypeScript SDKs support building local AI apps, scripts and workflows.

LM Studio can also connect local models to MCP servers, letting them use tools exposed through those integrations. For headless use on servers, cloud instances or CI, llmster runs LM Studio’s core without the desktop interface.

Main features

• Browse and download models: Find compatible models through the built-in catalog, which draws on Hugging Face, then manage downloaded files in LM Studio.
• Chat on your desktop: Load a local model and keep conversations in the app. The model you choose and the settings available depend on its format and compatibility.
• Ask questions about documents: Attach DOCX, PDF, and text files to a chat. LM Studio can provide shorter files as context directly, or use retrieval-augmented generation for longer documents.
• Connect software to a local model: Start a server for local applications and scripts. LM Studio offers its own REST API, plus OpenAI-compatible and Anthropic-compatible endpoints.
• Use local tools with MCP: Connect MCP servers to compatible models and workflows. Tool use depends on the model and the API path being used.
• Work from the terminal: Use the lms command-line tool to manage models, start the server, and run chat sessions. Python and TypeScript SDKs are available for developers.
• Deploy without the desktop interface: The llmster daemon runs LM Studio’s core in headless environments such as servers and CI.

Who should consider LM Studio?

LM Studio suits Linux users who want a graphical way to explore local models, chat with them, and make them available to other software. Developers can connect apps to local models through LM Studio’s API, manage them with the lms CLI, or run them headlessly with llmster. If you mainly want a terminal-oriented model service, check Ollama. If open-source desktop software is a priority, Jan is another option.

LM Studio’s main advantage is the bridge between model exploration and local development: you can try a model in the desktop app, then serve it to other tools. The trade-off is that model choice still depends on your computer’s available memory and processing hardware, while Linux support guidance is more cautious than the broad x64 and ARM64 labels suggest.