LM Studio
Desktop app for local LLMs — discover, download, chat
📋 Overview
LM Studio is a desktop application for running local LLMs — providing a polished interface for discovering models, downloading GGUF files, and chatting with any open-source LLM. Unlike Ollama (which is CLI-first), LM Studio provides a native desktop experience with model management, hardware monitoring, and a built-in chat interface. It supports GPU acceleration (CUDA, Metal, ROCm) and can run as an OpenAI-compatible server.
✨ Key Features
- •80K+ GitHub stars — most polished local LLM app
- •Model browser with one-click download
- •Automatic GPU detection (CUDA, Metal, ROCm)
- •OpenAI-compatible server mode
- •Hardware monitoring
- •Free for personal use
- •Hardware monitoring
- •Free for personal use
🎯 The Problem It Solves
Running local LLMs requires technical knowledge — downloading GGUF files, configuring GPU acceleration, managing models. LM Studio makes this accessible to non-technical users.
🔧 How It Works
LM Studio provides a model browser that searches HuggingFace for GGUF models — download with one click, and the app handles quantization selection based on your hardware. The chat interface supports multiple conversations, and the server mode exposes an OpenAI-compatible API. GPU acceleration is auto-detected.
🚀 Installation & Quick Start
Installation
Download from lmstudio.aiQuick Start
- Download LM Studio
- Browse models in-app
- Download a GGUF model
- Chat locally or start server
✅ Pros
- •Most polished local LLM app
- •One-click model download
- •GPU auto-detection
- •Server mode
- •Free personal use
- •Hardware monitoring
- •Free for personal use
❌ Cons
- •Commercial use requires license
- •Chat interface basic
- •Large models need high-end hardware
- •Commercial license required
💬 Practitioner Verdict
“LM Studio is the most polished local LLM desktop app — the model discovery and download experience is unmatched. The trade-off: commercial use requires a license, the chat interface is basic compared to Open WebUI, and very large models still need high-end hardware. For individuals who want to run LLMs locally, LM Studio is the default.”
Self-Hosted (Free)
Open source, MIT/Apache licensed. Run it yourself.
⭐ Star & Clone on GitHubFree forever. Your infrastructure, your data.
📊 Specifications
- Language
- TypeScript/Python
- License
- Proprietary (free personal)
- Platform
- Linux, macOS, Windows
- Supported Models
- REST API, CLI
💰 Pricing Reality
100% free for personal use. Commercial license required for business use (custom pricing).