LM Studio
A polished desktop application for discovering, running, and serving local models.
By LM Studio · reviewed 2026-08-07
Deep profile checked 2026-08-16 · LM Studio documentation and application releases
What LM Studio does
LM Studio combines model search, downloads, chat, runtime controls, and an OpenAI-compatible local server in a desktop interface. It is designed for people who want direct control over local models without managing a command-line stack.
Pricing model
Free desktop application
Position in the stack
Where LM Studio fits
LM Studio combines several layers that are separate in many local AI stacks: model discovery, downloads, a desktop chat interface, runtime controls, and a local API server. It is designed around one person's workstation rather than a shared infrastructure control plane.
Typical deployment flow
A practical sequence for evaluating LM Studio before making it part of a permanent stack.
- STEP 1
Install the desktop app
Use a supported Mac, Windows, or Linux machine and allow the application to inspect available compute resources.
- STEP 2
Download a compatible build
Select a GGUF or MLX model whose size and quantization leave enough memory for context and runtime overhead.
- STEP 3
Chat or start the API
Use the integrated interface or enable the local server for OpenAI-compatible development tools.
Best for
- Exploring local models from a desktop UI
- Apple Silicon and consumer workstation use
- Local API development without server administration
Not the right layer for
- Shared multi-user deployments
- Distributed production inference
System fit
Requirements and hardware notes
Application requirements are separate from the memory needed by the selected model and context window.
Check your hardwareMinimum baseline
- Apple Silicon Mac, supported Windows system, or supported Linux system
- macOS 13.4 or newer on Apple Silicon
- A model small enough for available memory
Recommended baseline
- 16GB or more RAM
- macOS 14 or newer for MLX models
- At least 4GB dedicated VRAM on Windows when using a discrete GPU
Hardware-specific notes
- Intel Macs are not currently supported.
- x64 Windows requires AVX2.
- GGUF runs through llama.cpp; Apple Silicon can also use MLX artifacts.
Installation and deployment paths
Choose one path that matches the number of users and the level of operations you can maintain.
Data boundary
What stays local and what may leave
- Local data
- Local model chat and document workflows can run on the machine; selected cloud providers or remote endpoints change that boundary.
- Network dependencies
- Model discovery and downloads use remote model repositories. Local inference can continue offline after required assets are present.
- Accounts and telemetry
- Review application settings and privacy documentation before using sensitive documents or optional online integrations.
License checkpoints
- LM Studio is free to use under its own terms; downloaded model licenses remain independent.
- Record the exact model repository and quantization author for reproducible reviews.
Operational checkpoints
- Use the runtime manager to keep llama.cpp or MLX runtimes explicit.
- A loaded model reserves memory until unloaded.
- Bind the API to a trusted interface and add access controls before LAN exposure.
Capabilities
Capabilities refer to the tool's application layer. Hardware and model support can still depend on a connected inference engine.
Deployment and compatibility
- Deployment
- Desktop / local
- Platforms
- macOS · Windows · Linux
- Hardware backends
- CPU · Apple Silicon · NVIDIA GPU · AMD GPU
- Install methods
- Desktop application
Models and integrations
- Model formats
- GGUF · MLX
- Common integrations
- OpenAI-compatible clientsMCP servers
Strengths
- Accessible model discovery and configuration
- Integrated chat and local server
- Supports GGUF and MLX workflows
Limitations
- Closed-source desktop product
- Less suitable for unattended server and cluster operations
Planning checklist
Before you choose LM Studio
Answer these questions with the exact models, hardware, users, and data you expect to operate.
- Is the target machine a personal workstation rather than a shared server?
- Do you need GGUF, MLX, or both model ecosystems?
- Will other devices need authenticated network access to the API?
- Do you require automated startup and operational monitoring?
LM Studio FAQ
Is LM Studio open source?
The desktop application is proprietary. It is free to use under its published terms, but it should not be treated as an open-source runtime.
Can LM Studio expose a local API?
Yes. Its local server supports OpenAI-compatible workflows, making it useful for development tools that expect that API shape.
Is LM Studio suitable for a team server?
It can serve local requests, but its primary strength is a desktop workflow. Dedicated serving platforms are generally better for multi-user operations and distributed deployments.
Official sources
Use these links to confirm current compatibility and installation requirements.
Related tools
An open-source desktop chat application and local API for running models privately on a personal computer.
A private desktop chat application and Python SDK for running GGUF language models on everyday computers.
A compact GGUF runtime and web interface focused on local text generation, roleplay, and story workflows.
Comparisons featuring LM Studio
Use a reviewed comparison when the choice is between two adjacent tools.
Ollama and LM Studio both make local models approachable, but they optimize for different workflows: an API-first runtime versus a desktop-first graphical application.
Jan and LM Studio both provide approachable desktop model discovery, chat, and local APIs, with the largest distinction being open-source licensing versus a proprietary polished product.
Execution evidence
Known working recipes using LM Studio
Recipes connect hardware, a model artifact, tools, settings, verification, and a reportable result.