Getting started with Ollama

What Ollama is, how to install it, and the handful of commands that cover everyday use.

What Ollama is

What: An open-source tool that downloads open-weight models, runs them on your CPU or GPU, and serves them through a CLI and a local HTTP API on port 11434.
Why: Private, offline and free per token: prompts never leave the machine, which suits sensitive data, prototyping and learning.

Watch out

Expecting a laptop model to match the largest hosted models on hard reasoning; pick tasks that suit small models.

Installing

What: Desktop apps for macOS and Windows, a one-line install script for Linux that sets up a systemd service, and an official Docker image.
Why: After install a background server is running, so the CLI and any app can use models straight away.

Shell

curl -fsSL https://ollama.com/install.sh | sh
ollama --version
systemctl status ollama
curl http://localhost:11434/api/version

ollama run

What: Downloads the model if needed, loads it and opens an interactive chat, or answers a single prompt passed as an argument.
Why: The fastest way to try a model and judge whether it suits your task.

Shell

ollama run llama3.2
ollama run qwen3:8b "Explain a mutex in two sentences."
cat notes.md | ollama run llama3.2 "Summarise this:"
ollama run llama3.2 --verbose

Chat session commands

Also called /set, /show, /bye

What: Inside ollama run, slash commands change settings for the session: /set parameter num_ctx 8192, /set system "...", /show info, /save name, /clear and /bye. Triple quotes start multi-line input.
Why: You can experiment with parameters and a system prompt interactively, then save the result as a new model.

pull, list, rm and show

What: pull downloads or updates a model, list shows downloaded models and their sizes, rm deletes one, and show prints its architecture, parameters, context length, template and licence.
Why: Models are gigabytes each; managing them keeps disks from filling up.

Shell

ollama pull gemma3:4b
ollama list
ollama show gemma3:4b
ollama show gemma3:4b --modelfile
ollama rm gemma3:4b

ps, stop and keep-alive

Also called keep_alive, OLLAMA_KEEP_ALIVE

What: ollama ps lists loaded models, their memory use and whether they run on GPU, CPU or both. Models unload after an idle period (five minutes by default); ollama stop unloads one now.
Why: Loading a large model takes seconds, so keep it loaded while working and free memory when done.

Shell

ollama ps
ollama stop llama3.2
OLLAMA_KEEP_ALIVE=1h ollama serve

The server and OLLAMA_HOST

What: ollama serve runs the server, normally started for you by the app or service. OLLAMA_HOST sets its address, 127.0.0.1:11434 by default, and also tells the CLI which server to talk to.
Why: Pointing the CLI at another machine lets a laptop use a GPU box on the network.

Shell

OLLAMA_HOST=gpu-box.lan:11434 ollama list
OLLAMA_HOST=gpu-box.lan:11434 ollama run llama3.2

Where models are stored

Also called OLLAMA_MODELS

What: Models are stored as content-addressed blobs under ~/.ollama/models, or in the service user's home on Linux. OLLAMA_MODELS moves them to another disk.
Why: Large models fill a small system drive quickly; layers shared between variants are stored only once.

Go deeper