LM Studio

What is LM Studio?

LM Studio is a desktop app for discovering, downloading, and running local and open-source Large Language Models (LLMs). It supports any ggml-compatible model available on Hugging Face, works across multiple operating systems, and uses your computer’s GPU when available to speed up model execution. Models run entirely offline, and you can interact with them through the built-in Chat UI or a local server with an OpenAI-compatible API.

Who created LM Studio?

LM Studio was built by a growing team. More details about the company are available on their careers page.

What is LM Studio used for?

• Experimenting with LLMs: Run local and open-source models on your own machine.
• Finding models: Discover and download ggml-compatible models directly from Hugging Face repositories without leaving the app.
• Model variety: Supports a wide range of ggml models, including Llama, MPT, and StarCoder, such as Llama 2, Orca, Vicuna, and NousHermes.
• Chat or server: Interact via the Chat UI or an OpenAI-compatible local server.
• Offline use: Once models are downloaded, no internet connection is required.
• GPU acceleration: Uses your GPU for faster model execution.

Who is LM Studio for?

LM Studio is aimed at software developers, data scientists, and researchers who want to experiment with LLMs locally. The interface is straightforward to configure, making it easy to go from downloading a model to running inference.

Getting started

LM Studio runs on M1/M2/M3 Macs, Windows PCs with a processor supporting AVX2, and as a Linux beta. After installing the app, browse and download a ggml-compatible model from Hugging Face, then use it through the Chat UI or serve it via the OpenAI-compatible local server.