Mindket
English
English简体中文
Sign inSign up
StoreMCPSkillsCalculatorLearnPriceBlog
Discover
Research308Writing59Business95Legal13Finance & accounting33Marketing & SEO172Developer1,379Productivity382Education81Companions14
Help and support
Language
English简体中文
IntermediateFree

Local AI course: run AI models on your own computer

Mindket editors · Updated · 4 lessons · 43 min

A black banner titled "Local AI" over the line "Run models on your own computer".

This local AI course shows how to run open AI models on your own computer, private and offline once they are downloaded. You install Ollama and run a model, call its API from code, use LM Studio and its local server, then work out how much memory a model needs. Every step follows the tools' documentation as it stood on 5 November 2026.

What you'll be able to do

  • Install Ollama, pull a model, then run, list and remove models from the terminal.
  • Call the Ollama API with curl and Python, stream replies and build a custom model with a Modelfile.
  • Download and chat with models in LM Studio, then serve them through its OpenAI-compatible server and the lms command.
  • Work out how much memory a model needs from its parameters, its quantization and its context length.
All courses

Who it's for

For people comfortable with a terminal who want to run AI models on their own computer, private and offline. You need a Mac, Windows or Linux computer, ideally with a graphics card or an Apple silicon chip, and Python for the code examples.

Lessons

  1. 1Ollama tutorial: run AI models on your own computerAn Ollama tutorial for Windows, macOS and Linux: install it, download and chat with a first local model, then list, stop and remove models and move their files.11 min
  2. 2Ollama API: call a local model from your codeCall a local model from code with the Ollama API: /api/generate and /api/chat with curl and Python, streaming, the OpenAI-compatible endpoint and a Modelfile.11 min
  3. 3LM Studio tutorial: run local models in a desktop appAn LM Studio tutorial: check your hardware, get a model from the built-in search, chat with it and your files, then run its OpenAI-compatible server and lms.11 min
  4. 4Local LLM hardware requirements: how much VRAM a model needsLocal LLM hardware requirements, worked out: how parameters and GGUF quantization set a model size, how much VRAM it needs, and what CPU, GPU or a Mac changes.10 min

Frequently asked questions

Do I need a GPU for this course?

No, but it helps. Ollama and LM Studio can run models on the processor from system RAM, only slower. Ollama recommends 8 GB of VRAM, or unified memory on a Mac, for its quickstart model, and lesson 4 shows how to size any other model.

Does the course use Ollama or LM Studio?

Both. Lessons 1 and 2 use Ollama from the terminal and through its API, and lesson 3 uses the LM Studio app, its server and its lms command. Both run GGUF models through llama.cpp, so the sizing in lesson 4 applies to either.

Is running a model locally private?

The model runs on your machine, so the prompts and files you give it stay there. Downloads, app updates and any cloud features you turn on still use the internet, and a server you share on your network is open to other devices.

What should I learn after this course?

The Hugging Face course is next. It starts with the Hub, where the models and GGUF files used in this course are published.

Share this course

Give a local model tools

Browse MCP servers on Mindket with their setup notes, then add the ones you trust to LM Studio or another app that hosts your model.

Browse MCP servers

All courses · All articles