Install Ollama and a model

Install Ollama, start its local server and download at least one text model. A downloaded model and running server are both required.

Connect ActionClip

  1. Open Settings → AI → Configure… → Ollama Local.
  2. Review Local server address, normally http://localhost:11434, then choose Connect.
  3. Select a downloaded Provider default model.
  4. Set Ollama Local as the Settings default, or choose it in an AI Prompt action’s provider/model selectors.

Choose a downloaded model

The model selector lists available downloaded models. An action can inherit the provider’s default model or select its own. If the server has no local models, download one in Ollama and reconnect.

What counts as local

Loopback addresses such as localhost, 127.0.0.1 and IPv6 loopback stay on your Mac. Ollama Local accepts local loopback servers. Configure a remote server through the compatible API connection; use Ollama Cloud for the cloud service. Local Ollama does not require Allow online actions.

Troubleshooting

Check the running server, local address/port and downloaded model. A model that runs in Ollama must also be available through that server. Disconnect returns the connection to its empty setup state.

Back to all guides