SettingsΒΆ

The Settings tab of the GeoAgent panel. The i buttons next to each option explain it.

The Settings tab with Ollama selected

Setting

What it does

Provider

Which LLM service to use: Ollama, OpenAI, Gemini or Anthropic. See LLM providers.

Temperature

0 to 1 (default 0.8). Lower values make answers more consistent, which suits processing tasks. Not sent to Anthropic models.

Maximum Token

Upper limit on the length of each model response (1,000 to 100,000; default 5,000).

After a processing run

Open each processing run in the Model Designer, ask to save it as .model3, or just show the result (default). See Model export.

Model Name / API Key

For OpenAI, Gemini and Anthropic: the model to use and your key.

Ollama Base URL / Ollama Model Name

For Ollama: the server address (default http://localhost:11434) and the model.

Check / Install Dependencies

Installs missing Python packages. See Installation.

Save Settings

Keeps all of the above across QGIS restarts, with a separate API key per provider.