Lesson 3 of 7

Tools to run them

6 min read

Running a model yourself sounds like a job for engineers. So how does a beginner actually get one going?

One app, three steps

You don't build anything — a runner app does it all. Ollama and LM Studio are the popular ones: you pick a model, they download ('pull') it once, and then you chat with it in a normal window. LM Studio is fully point-and-click; Ollama is a tiny app many other tools plug into. Either way it's download-once, then chat — no coding.

A runner app (Ollama, LM Studio) makes it download-once, then chat — no command line needed.

After the download, it's just chat

The first pull takes a while and some disk space — the model is a big file. After that it's instant and offline: the model sits on your drive and answers straight from your machine, no account and no connection. You can keep several models and switch between them, the same way you'd switch apps.

Pull a model once; from then on it runs offline, instantly, straight off your disk.

Start with a small model to see it work — it downloads fast and runs smoothly. You can always pull a bigger one once you know your machine handles it.