LM Studio

lmstudio.ai

LM Studio is a free desktop app for running open-weight LLMs like Llama, Qwen, and DeepSeek locally on your own computer, with a built-in model browser, chat UI, and an OpenAI-compatible local API server.

Overview

LM Studio is a desktop application that downloads and runs open-weight large language models entirely on your own hardware — Windows, macOS, and Linux. Built by Element Labs, it wraps the same llama.cpp and MLX inference engines that power most of the local-LLM ecosystem behind a clean, click-driven GUI: a built-in model browser pulls from Hugging Face, one click downloads and loads a GGUF or MLX model, and a ChatGPT-style chat window lets you talk to it immediately. Nothing leaves your machine, which is the entire point. Since July 2025 the core app has been free for both personal and commercial use, with no license form or registration. A toggle exposes any loaded model as an OpenAI-compatible server at localhost:1234, so your existing code and tools can call a model running on your laptop with zero changes. Recent builds added tensor-parallel multi-GPU, speculative decoding, and a companion mobile app that streams from your desktop.

Key Features

  • Built-in model browser with one-click downloads from Hugging Face
  • ChatGPT-style chat window with per-model inference settings
  • OpenAI-compatible local API server at localhost:1234
  • Dual engine: llama.cpp (GGUF) on NVIDIA/AMD/Intel/CPU, MLX on Apple Silicon
  • Tensor-parallel multi-GPU and speculative decoding for faster generation
  • JS and Python SDKs plus a headless llmster daemon for server-only deploys

Pricing

PlanPriceFor
Free$0Personal and commercial use, all features, local API server
EnterpriseCustomSSO, model/MCP gating, private collaboration, advanced admin

Comparison

Compared to Ollama, LM Studio trades the command line for a polished GUI and visual model management — better for newcomers who don’t want a terminal, while Ollama stays the leaner choice for headless servers and automation. Against Jan, LM Studio offers deeper hardware optimization and a richer API story, though both are free local runners. If you specifically want a managed cloud RAG stack instead of local inference, Hugging Face is the better fit.

Compare alternatives

Side-by-side with the 3 closest alternatives.

ToolCategoryPricingVisit
LM Studio (this) code, productivityFree $0 Site ↗
Ollamacode, chatFrom $0/mo Site ↗
Janchat, productivityFree $0 Site ↗
Hugging Facecode, searchFree $0 · From $0.5/mo Site ↗
LM Studio Current

LM Studio is a free desktop app for running open-weight LLMs like Llama, Qwen, and DeepSeek locally on your own computer, with a built-in model browser, chat UI, and an OpenAI-compatible local API server.

codeproductivity
Free $0

A desktop tool to download and run open large language models locally with one command. Runs models fully offline on your machine Read our hands-on review

codechat
From $0/mo

An open-source ChatGPT-alternative that runs AI models locally on your computer. Clean desktop chat UI for local models Read our hands-on review and

chatproductivity
Free $0

The central hub for open ML — host and run a million-plus models, datasets, and Spaces, with serverless and dedicated inference.

codesearch
Free $0 · From $0.5/mo
Editor’s Review
4.6/5
Pros
  • +Free for both personal and commercial use with no license form or registration
  • +Polished GUI with one-click model discovery and downloads from Hugging Face
  • +OpenAI-compatible local server lets existing code call your own hardware with zero changes
Cons
  • Closed-source core, so you cannot audit or self-modify the app itself
  • Heavier RAM and VRAM footprint than lean command-line runners

LM Studio is the closest thing the local-AI world has to a just-works desktop app: download, click, chat, done — no terminal required. It is the right first stop for anyone who wants privacy-first inference without learning Docker or llama.cpp flags, though power users running headless servers at scale will still prefer Ollama.

See all reviews →