LM Studio is a powerful free desktop application for running open-source language models locally on your own hardware. It features a modern GUI, built-in model discovery from HuggingFace, GPU acceleration (NVIDIA, AMD, Intel, Apple), and an OpenAI-compatible local API server. LM Studio is free for both personal and commercial use and provides offline operation with full privacy.
License: Free for personal and commercial use (Proprietary)
Latest Version: 0.4.6 (February 27, 2026)
Website: lmstudio.ai
GitHub: lmstudio-ai
- 🖥️ Modern Desktop UI - Clean, intuitive interface with split view support
- 📥 Model Discovery - Built-in HuggingFace integration with search
- ⚡ GPU Acceleration - NVIDIA CUDA, AMD ROCm, Intel Arc, Apple Metal/MLX
- 🔌 OpenAI-Compatible API - Drop-in replacement for OpenAI API calls
- 📚 Document Attachments - RAG with local files (PDF, TXT, MD)
- 🎯 Speculative Decoding - Faster inference with draft models
- 🔧 Developer Mode - Advanced options and per-model settings
- 🤖 MCP Support - Model Context Protocol integration
- 🌐 LM Link - Connect to remote LM Studio instances (E2E encrypted)
- 💻 CLI Tool (lms) - Terminal-based model management and chat
- 🐍 Python SDK -
lmstudio package for Python integration
- 📦 TypeScript SDK -
@lmstudio/sdk for JavaScript/TypeScript
- 🖥️ llmster - Headless daemon for server/cloud/CI deployments
- Local LLM Development - Test and experiment with models offline
- Privacy-Focused AI - Fully local operation, no data leaves your machine
- API Backend - Local OpenAI-compatible server for applications
- Document Analysis - RAG with local PDFs and documents
- Model Benchmarking - Test different models and configurations
- Team Collaboration - Share models via LM Link or Hub organization
- Enterprise Deployment - Internal AI tools with SSO (Enterprise plan)
| Component |
Technology |
| Backend |
C++, llama.cpp |
| Frontend |
Electron (TypeScript) |
| Inference Engine |
llama.cpp, MLX (Apple) |
| API |
OpenAI-compatible REST API |
| CLI |
Node.js (lms) |
| SDKs |
Python, TypeScript (MIT licensed) |
| Component |
Minimum |
Recommended |
| OS |
Windows 10, macOS 11, Linux |
Latest OS version |
| RAM |
16 GB |
32-64 GB |
| GPU VRAM |
4 GB |
12-24+ GB |
| Disk |
20 GB |
100+ GB (NVMe SSD) |
| GPU Support |
NVIDIA, AMD, Intel, Apple |
Latest GPU drivers |
- ✅ Free for personal and commercial use (since July 2025)
- ✅ Desktop apps for Windows, macOS, Linux
- ✅ GPU acceleration (CUDA, ROCm, Metal, MLX)
- ✅ OpenAI-compatible API server
- ✅ Python and TypeScript SDKs (MIT licensed)
- ✅ CLI tool (lms) included
- ✅ llmster headless daemon available
- ✅ LM Link for remote connections
- ⚠️ Proprietary application (not open-source)
- ⚠️ Enterprise features require paid plan
Any questions?
Feel free to contact us. Find all contact information on our contact page.