Linux · System Tools
LM Studio
A desktop app for running local LLMs offline with a chat UI, model manager, and OpenAI-compatible API server.
- License
- LicenseRef-proprietary=https://lmstudio…
- Latest version
- 0.4.23-1 (August 28, 2026)
- Developer
- Element Labs, Inc.
- Installs
- 12,600 (Flathub, 30 days)
What is LM Studio?
LM Studio is a cross-platform desktop application developed by Element Labs, Inc. It allows users to run large language models locally without relying on cloud services. The software has matured quickly, offering a unified interface for discovering, downloading, and managing GGUF and MLX model formats from Hugging Face. Recent versions include headless mode for servers and MCP integration for broader AI workflows.
This tool is ideal for developers, privacy-conscious users, and hobbyists who want to experiment with open-weight models like Llama, Mistral, or Qwen. It serves as a local alternative to cloud-based assistants, giving you full control over the hardware resources allocated to inference. You do not need to be a backend engineer to start, but the developer tooling makes it useful for integration projects.
Compared to Microsoft Copilot, LM Studio feels like a local terminal with a friendly face. There is no subscription or data upload, but you are responsible for selecting the right model size for your hardware. The interface is clean and ChatGPT-like, but the power lies in the backend controls for GPU offloading and memory management, which Copilot abstracts away completely.
Best for: Users who want to run private, offline AI assistants and have the hardware to support it.
Key features of LM Studio
- Run open-weight LLMs offline without cloud dependency
- Browse and download GGUF/MLX models from Hugging Face
- Built-in ChatGPT-like chat interface
- OpenAI-compatible REST API server for app integration
- Document Chat (RAG) for private, offline document Q&A
- Hardware-aware optimization with GPU offloading controls
- CLI, Python, and TypeScript SDKs for developer use
- Headless mode for running on servers or CI environments
Pros
- Total privacy: prompts and data never leave your machine
- No subscription fees or token limits
- Easy model discovery and management in one place
- OpenAI-compatible API makes integration straightforward
Cons
- Requires significant local hardware (RAM/GPU) for usable performance
- Proprietary license, not open source
- You must manually select and download models; no automatic model selection
- Performance depends heavily on your specific hardware configuration
Windows programs LM Studio replaces
How to install LM Studio on Linux
flatpak install flathub ai.lmstudio.lm-studioTips for Windows users
Unlike Copilot, you must first download a model file (GGUF format) that fits your RAM and VRAM. Start with a smaller model if you have limited hardware. There are no account logins or cloud sync; your chat history and settings are stored locally. If you were using Copilot for quick web searches, note that local models do not have real-time internet access unless you configure a RAG pipeline with your own documents.
Recent releases
- 0.4.23-1 —
- 0.4.18-1 —
- 0.4.17-4 —
- 0.4.16-2 —
LM Studio: frequently asked questions
Does LM Studio work offline?
Yes, once a model is downloaded, it runs entirely on your local hardware without an internet connection.
What file format does LM Studio use for models?
It primarily uses GGUF files, which are quantized versions of large language models optimized for local inference.
Can I use LM Studio as an API for other apps?
Yes, it includes a built-in server that exposes an OpenAI-compatible REST endpoint, allowing you to point other applications at your local model.
Is LM Studio free to use?
The desktop application is free to download and use. You only pay for the compute resources (electricity and hardware) to run it.
Updated