The Models section under Define lets you configure the primary Large Language Model (LLM) used by an AI Agent.

The primary model:

• Acts as the default model for all agent tasks and applications
• Ensures consistent model behavior
• Simplifies centralized configuration management

Changes made at the AI Agent level automatically apply to all tasks that use the default configuration. Individual tasks can override these settings when required.

Configure the Primary Model

1. Open the required AI Agent in AI Agent Builder.

2. From the left navigation pane, click Define.

3. In the left pane, select Models.

   

4. Under Primary Model, configure the following settings. To configure advanced model parameters, turn on Additional Settings.

   • LLM Provider – Select the provider for the primary LLM. By default, the provider configured for the AI Agent is selected.

   • Model – Select the model to be used by the AI Agent. The default model configured for the AI Agent is preselected.

   • Max Tokens – Specify the maximum number of tokens that can be generated in a response.

   • Temperature – Control the variability of generated responses. Lower values produce more consistent responses, while higher values allow greater variation.

   • Frequency Penalty – Adjust the penalty applied to repeated words or phrases in the generated output.

   • Presence Penalty – Adjust the penalty applied to topics that have already appeared in the conversation.

   

5. Click Save in the bottom-right corner to apply the configuration.

Task-Level Model Overrides

Tasks inherit the primary model configuration by default.

Override Behavior

• You can configure a different:

  • LLM provider
  • Model
  • Model parameters

• When a task uses a custom configuration:

  • It is treated as a task-level override
  • AI Agent-level updates do not affect the overridden settings.

• Tasks without overrides continue to follow the default configuration.

​