Developing and experimenting with large language models (LLMs) often requires significant computational resources and complex setup processes. Many developers and researchers lack access to the hardware or expertise needed to effectively run and test LLMs locally. This creates barriers to entry and slows down innovation in the field.

LM Studio is a desktop application designed to simplify the process of running and experimenting with LLMs on personal computers. It allows users to discover, download, and run models such as Llama, Mistral, and Phi directly on their machines, entirely offline. The application provides a user-friendly chat interface for interacting with models and an OpenAI-compatible local server for deploying models in custom applications. LM Studio supports various model architectures and offers tools for managing local models and configurations, making LLM development more accessible.

Local LLM execution supporting architectures like Llama, Mistral, Phi, Gemma, StarCoder, and DeepSeek.

Model discovery and download from Hugging Face repositories directly within the app.

Built-in chat interface for interacting with local models.

Local server with OpenAI-compatible endpoints for API access.

Offline operation ensuring data privacy and security.

Document chat enabling interaction with local documents using Retrieval-Augmented Generation (RAG).

Support for GGUF Llama.cpp and MLX models.

Cross-platform compatibility with macOS, Windows, and Linux.

The primary users are developers, researchers, and hobbyists who want to experiment with LLMs locally without relying on cloud services or complex setups.

LM Studio is free for personal use; commercial use requires contacting the company for licensing.