Model configuration and extended thinking options

The Prompt Catalog includes the Model Config side panel for advanced model settings.

The panel includes these options:

  • Max response length: Specify the maximum number of tokens that the model can generate in a response.
  • Thinking switch: Enable or disable extended reasoning. The switch is displayed when the selected model supports reasoning, as indicated by the reasoning flag in the GET/models response.
  • Temperature percent: Control the randomness of the model output.
  • Top P percent: Control output diversity through nucleus sampling. The field is not available for Claude 4.5 models and later.

After the response is generated, these response metrics are available:

  • Response time
  • Total tokens