Home / Library / Help / Set up a runner / Ollama
Guide
Set up Ollama
Ollama runs a language model on your own machine and serves it over a local API that mi·do·na can talk to. Nothing leaves your computer. It's command-line first and pleasantly minimal - one command pulls a model, and the server runs itself in the background. This is the short version; Ollama's own docs cover the rest.
Get a model running
Download & install Ollama
Grab the build for your OS from ollama.com/download. macOS and Windows have installers; on Linux it's a one-line install script. Once it's running you'll have the
ollamacommand in your terminal.Pull a model
Browse the library at ollama.com/library, then pull one from your terminal:
Terminalollama pull llama3.1:8b
If you're unsure, start with something in the 7-8B range that fits your RAM. Bigger models are smarter but heavier.
It's already serving
Ollama runs a local server in the background as soon as it's installed, listening on port
11434. It exposes an OpenAI-compatible API athttp://localhost:11434/v1. Confirm what you've got withollama list.Allow browser access (CORS)
Because mi·do·na runs in your browser, Ollama has to permit browser requests. Set the
OLLAMA_ORIGINSenvironment variable to allow them, then restart Ollama. On macOS:Terminallaunchctl setenv OLLAMA_ORIGINS "*"
On Windows or Linux, set
OLLAMA_ORIGINS=*in your environment and restart the app. To be stricter, use mi·do·na's exact origin instead of*.
Next: connect mi·do·na
With a model pulled and origins allowed, Ollama is ready. Open the connection
panel in mi·do·na, point it at http://localhost:11434/v1, and load
models. See Connect a model.
Served over HTTPS and can't reach localhost? The browser
can block an HTTPS page from talking to http://localhost - most
often in Safari. See
Reaching a local model over HTTPS
.
This is a quick-start. For anything deeper, see the official Ollama documentation at github.com/ollama/ollama.