AI Tools

Running AI Locally with LM Studio

Download, install, and chat with an AI that runs entirely on your computer

Normally when you use AI, your messages travel to a company's servers and back. LM Studio lets you skip that entirely — it downloads an AI model to your computer so everything runs locally. No internet needed, no account required, and your conversations never leave your machine.

Why would I want this?

Privacy — Your conversations stay on your computer. Period.
Free — No subscription, no usage limits, no API costs
Offline — Works without internet (after you download the model)
Learning — A great way to understand how AI models actually work

Local models are less powerful than Claude or ChatGPT. Think of them as a handy assistant for everyday tasks, not a replacement for the best cloud models.

Can my computer handle it?

AI models need a decent amount of memory (RAM) to run. Here's how to check what you have:

Mac: Click the Apple menu (top-left) → "About This Mac." Look for "Memory" — you want at least 8 GB.
Windows: Press Ctrl + Shift + Esc to open Task Manager → click "Performance" → "Memory." Check the total.

Got 8 GB of RAM? You can run small models. 16 GB? You're in great shape for most models. Apple Silicon Macs (M1/M2/M3/M4) are especially good at this.

Step 1: Install LM Studio

Go to lmstudio.ai in your browser
Click the download button — it detects your operating system automatically
Mac: Open the downloaded .dmg file and drag LM Studio to your Applications folder
Windows: Run the downloaded .exe installer and follow the prompts
Open LM Studio. No account or sign-up needed.

Step 2: Download a model

A "model" is the AI brain itself — a large file (usually 2–5 GB) that LM Studio runs on your hardware.

Click Discover in LM Studio's sidebar
Search for Llama 3.1 8B — it's a great starter model that runs well on most machines
Click the Download button and wait. It's a big file — this might take a few minutes on slower internet.
If your computer struggles, try a smaller model like Gemma 2 2B or Phi-3 Mini

The numbers in model names (like "8B") refer to billions of parameters — roughly how complex the model's "brain" is. Bigger = smarter but slower and more resource-hungry.

Step 3: Start chatting

Click the Chat tab in the sidebar
Select the model you just downloaded from the dropdown at the top
Type a message in the chat box and hit Enter
The first message might take 10–30 seconds — the model needs to load into memory. After that, responses are faster.

Tips once you're up and running

If responses are too slow, try a smaller model
Look for models labeled Q4 — they're compressed to run faster while keeping most of the quality
LM Studio can also run a local API server, which means other apps on your computer can talk to your local model (useful if you get into coding)

Try this now

Install LM Studio, download Llama 3.1 8B, and ask it to explain a topic you're studying. Then ask Claude or ChatGPT the same question. Compare the answers — you'll get a feel for what local models do well and where cloud models still have the edge.

← Guide 1 All Guides Guide 3 →