DOA
Help me choose

Local AI › Local AI: frequently asked questions

Local AI: frequently asked questions

Checked on 13 Oct 2026. Versions change every week: check the official site before installing.

Is local AI free?

The tools in this guide can be used for free: Ollama, GPT4All and llama.cpp are MIT-licensed, Jan Apache 2.0, ComfyUI GPL-3.0, Open WebUI has its own licence and LM Studio is free under its terms of use. What you pay for is the computer and the electricity. (source)

Does it work offline?

Yes, once the model is downloaded. LM Studio documents that it can run fully offline. You need internet to install the tool and download the models. (source)

Does my data leave my computer?

With a model that runs on your computer, what you type is processed there. The exception is models that run in the cloud, such as those marked «-cloud» in the Ollama library. (source)

How much memory do I need?

At least the model's file size, plus some more for the context and the system. The models-by-RAM page lists the sizes published by Ollama and Hugging Face for 8, 16, 24 and 32 GB or more.

Which tool should I start with?

LM Studio if you prefer an app with windows, or Ollama if you don't mind typing a command. Both run on Windows, macOS and Linux.

Is GPT4All still active?

Its latest release on GitHub is v3.10.0, from 25 February 2025. There have been no new releases since, so we mark it as low activity. (source)

Is it as good as ChatGPT or Claude?

In general, models that fit on a regular computer fall short of the top cloud models at complex reasoning. For writing, summarising, translating or help with code they are usually enough.

More detail in the tools and the models by RAM.

Keep reading