LM Studio
Run open models locally — private, free, offline
Overview
LM Studio is the friendliest way to run open-weight models (Llama, DeepSeek, Qwen, Gemma) on your own machine: a clean chat UI, one-click model downloads, an OpenAI-compatible local server, and full offline privacy. Essential for sensitive work and zero-cost experimentation.
Features
How to Get Started
Download from lmstudio.ai (macOS/Windows/Linux), search a model (try a small Qwen or Llama first), download a quantization that fits your RAM, and chat — or enable the local server and point any OpenAI-SDK app at it.
Pricing
Free
View pricing detailsFAQ
What hardware do I need?
8GB RAM runs small models; 16-32GB (or Apple Silicon) runs genuinely useful 7-32B models. Bigger models need serious GPUs.
LM Studio vs Ollama?
Ollama is CLI-first for developers; LM Studio adds a full GUI and model browser — many use both.
Can local models replace Claude/GPT?
For private drafts, classification and offline work — increasingly yes. For frontier reasoning, not yet.
Read AI & Design Articles
Stay up to date with the latest trends and tips in design and AI
Learn AI Skills
Discover practical skills for using AI tools effectively