Local models in one command
Ollama is a free tool for running open LLMs such as Llama on your own computer. It collapses the fiddly business of obtaining, managing, and running models into a single command, and has become the standard entry point for local LLMs.
What you get
- Simple: typing ollama run followed by a model name handles download and startup
- Free: the tool and the models cost nothing — your hardware is the only constraint
- Private: conversations never leave the machine
- Cross-platform: Mac, Windows, and Linux, with a chat application available
Running an LLM locally used to require specialist knowledge. This removed most of that barrier.
Which models run
It supports many published models, including the Llama family, open models from labs worldwide, and quantized lightweight versions. Starting with a small model matched to your memory and GPU is the realistic route. Being called from other applications is common too, which makes it the base layer for local AI.
If you want to try it
If you do not want your input leaving the machine, want to experiment with LLMs at no cost, or want to understand how these systems work, it is worth the time. Installing from the official site and running a small model takes under half an hour. It is not as capable as frontier cloud models, but having an AI running on your own computer deepens your understanding considerably. LM Studio is the equivalent for people who prefer a graphical interface.
Ollama also increasingly supports multimodal models that understand images, not only language models. What is possible on your own hardware is expanding faster than most people expect.