Set Up Ollama

All OpenSymbolicAI tutorials run against a local model by default. No API key, no sign-up, no usage bill. The only requirement is Ollama, a runtime that serves models locally. Setup takes about five minutes.

1. Install Ollama#

Follow the instructions on the official Ollama download page for your platform. Once installed, Ollama runs as a background service.

Verify it's alive:

ollama --version

2. Pull the recommended model#

The tutorials use qwen2.5-coder:7b, a compact coding model that fits on most laptops (around 4 GB of disk space):

ollama pull qwen2.5-coder:7b

The first pull takes a minute or two. Subsequent starts are instant because the model is cached locally.

3. Confirm it responds#

Run a quick smoke test from the terminal:

ollama run qwen2.5-coder:7b "reply with the single word: ready"

You should see:

ready

If the command hangs or errors, check the note below.

4. How tutorials use Ollama#

Each tutorial creates an LLMConfig pointing at the local Ollama service:

from opensymbolicai.core import LLMConfig

llm = LLMConfig(provider="ollama", model="qwen2.5-coder:7b")

That config gets passed to the agent. No API key, no environment variable.

You're ready#

Ollama is running and the model is pulled. Head to the first tutorial: Hello, OpenSymbolicAI.