LM Studio setup guide for running an LLM locally with privacy

Analyst running a local LLM privately on a Microsoft Surface Laptop in a closed office

LM Studio is a desktop app with a chat interface that lets you get and run a large language model (LLM) on your own computer, without sending your content to external servers on the web. An LLM is an artificial intelligence (AI) model trained on large volumes of text. Once a model is downloaded, it works offline.

What you’ll take from this

You will install LM Studio and chat with a model that runs on your own computer. To bring that skill into your job, the Certificate Program in Artificial Intelligence and Innovation from Founderz lists LM Studio and Ollama among its tools.

  • Choose a model that fits your RAM
  • Start a local API server
  • Decide between LM Studio and Ollama
  • Check every local answer before you rely on it

What is LM Studio and what does it do?

LM Studio is a desktop application where you search for open language models, download them and run them locally, with no terminal required. The model generates text, a step called inference, on your computer’s processor or graphics card. A cloud chatbot sends every prompt to a provider’s servers instead, so local running gives you more control over your files and context.

A typical use is a manager summarizing meeting notes that should not leave the laptop. Others draft emails, test code snippets or create a prototype against a local server.

How do you install and use LM Studio on your computer?

Installing LM Studio needs no coding. You get the installer from the official site, lmstudio.ai, and follow four steps. After the model file is on your disk, chatting works without an internet connection.

  1. Install the application.
  2. Search for a model in the built-in browser and download it.
  3. Load it in the chat tab.
  4. Type your first question.

Turn off your wifi, then paste this test prompt:

Summarize this meeting transcript in five bullets and list the three decisions still unresolved: [paste your text]

If the model answers with the network off, it runs locally. Read the answer against your original text, because a small model can miss a decision or invent one.

Which model can you download and what hardware do you need?

You can download open models available in GGUF format, such as Llama, Gemma from Google or DeepSeek, as long as the file fits in your RAM or GPU memory with room to spare. If a model fails to load, pick a smaller model. LM Studio’s system requirements recommend at least 16GB of RAM on Windows and advise smaller models and modest context sizes on 8GB Macs.

For example, a 5 GB file on an 8 GB laptop that is also running a browser and email will struggle. A smaller model has a better chance of loading. Start small, then move up.

Open models keep closing the gap with paid ones. According to the Stanford AI Index (2025), the performance difference between open-weight and closed models fell from 8% to 1.7% on some benchmarks in a single year. The same report says performance on MMMU rose 18.8 percentage points and on the coding benchmark SWE-bench 67.3 points in one year. Models change fast, so verify every answer.

Which is better for running LLMs locally, Ollama or LM Studio?

LM Studio is better if you prefer windows and a built-in model search. Ollama fits if you are comfortable in the terminal. Both appear in the tools covered by the Founderz program, so try each on the same task before you choose.

CriterionLM StudioOllama
Main useChat in a desktop windowCommand line
Model downloadBuilt-in searchTerminal command
Best forNon-technical professionalsPeople at ease with a terminal

For a related look at organizing AI steps into a routine, see our guide to an AI creative workflow.

How do you turn on the LM Studio local API server?

Load a model in LM Studio and start the local server from the app. Other applications then connect to it using the OpenAI API format, so existing code or tools can point at your computer instead of a cloud service.

Features such as MCP connections, a command-line interface (CLI) and agent apps, where an agent chains several steps on its own, change quickly, and new versions arrive often. Confirm them in the official LM Studio documentation and support pages before you plan a workflow around them. Treat the server like any internal service and keep it on your own machine unless your IT team approves wider access.

How do you keep local AI use private and safe at work?

Running a model locally reduces exposure, but it does not remove your duties. Use this checklist before you rely on it for work tasks.

  • Confirm the model file fits your memory and loads without errors.
  • Run the offline test with your wifi off.
  • Read the model’s license before using it in a company.
  • Keep personal data out of prompts unless you have a lawful basis under the GDPR.
  • Verify names, numbers and quotes against your source.

Frequently asked questions

These are the questions people ask most before they install LM Studio.

Where do you download LM Studio and for which systems?

Download it from lmstudio.ai, the developer’s official site, with versions for Windows, macOS and Linux. Avoid third-party mirrors, because an installer from an unknown source can carry malware. Check that the version on the download page matches your operating system before you install.

Is your data private with LM Studio?

With a downloaded model, processing happens on your machine and your text is not sent to external servers, which is the main privacy benefit. Searching for models and fetching them still needs a connection. If you handle personal data, Regulation (EU) 2016/679 (GDPR) still applies.

Does LM Studio cost anything?

Pricing and license terms can change, so check the official lmstudio.ai site and its terms on the day you install, especially for commercial use at work. Each model also carries its own license, which you should read before using it inside a company.

Can you trust LM Studio?

Trust it as you would any software. Download only from lmstudio.ai, take models from publishers you recognize and read each license. Trust the output less than the app, because any LLM can state wrong facts with confidence. Check names, numbers and quotes against your source before sharing.

Further reading

These official sources cover the benchmark data and the data protection rule cited above.

If you want to apply local and private AI at work, request information about the Certificate Program in Artificial Intelligence and Innovation from Founderz. It is fully online, takes 3-4 months, includes 12 practical modules and ends with the Microsoft + Founderz Certification and official Microsoft badges.

Paul Delaney

Paul Delaney has been engineering AI prompts since the GPT-2 era, long before ChatGPT made prompting mainstream. Paul leads SEO, AEO, and GEO strategy at Founderz, improving how the school and its programs are discovered through traditional and AI-powered search. With more than 25 years of experience in education and digital growth, he has used AI daily since 2021 to support his commercial work.