Configure models and manage model lifecycle

AI Agent Studio now gives you more control over the large language model (LLM) that each workflow uses. From the LLM tab in Workflow Settings, you can select a model based on your workflow requirements and use model descriptions to compare capabilities and intended workloads. The model that you select remains associated with the workflow, which helps provide consistent and predictable results across development, testing, and runtime.

Models are grouped into basic and balanced categories:

• Basic: Supports cost-sensitive workloads such as summarization, question answering, retrieval-augmented generation (RAG), and lightweight reasoning.
• Balanced: Supports workloads that balance quality, latency, and cost, including RAG, tool use, multistep reasoning, and agent orchestration.

The Balanced category replaces the former Premium category. This terminology change doesn't affect model capabilities.

AI Agent Studio identifies deprecated models so that you can distinguish them from active models when configuring a workflow. When a model is deprecated, review the affected workflows and plan a transition to an alternative active model before the deprecated model is retired.

Explicit model selection gives your organization greater control over the models used by workflows. It helps prevent platform model updates from changing workflow behavior unexpectedly and supports more consistent results across development, testing, and runtime.

Steps to enable and configure

You must have access to use AI Agent Studio.

Access requirements

Access Requirements for AI Agent Studio