1. How do I use AI Agent Studio?
  2. Add Your LLM

Add Your LLM

In addition to the LLMs provided by Oracle that are available for selection when creating an agent, node, or agent team, you can also use other LLMs. You can use the additional LLMs available in AI Agent Studio by adding the credentials for it. If the model you want to use isn't available in the list of models for the supported provider, you can submit a request through My Oracle Cloud Support to get the model added.
Important: Oracle updates the list of Bring Your Own LLM options on a monthly basis and may remove models that have been deprecated, based on information provided by their respective model providers. You're responsible for obtaining and managing all required third-party provider credentials, associated service costs, and custom model configurations. You must also monitor model lifecycle changes, keep configurations up-to-date, and complete any required migrations to supported model versions within the timelines specified by the model provider.
You can add an LLM from any of these supported providers:
  • OpenAI (through Microsoft Azure)
  • Google Gemini (through Google Vertex AI)
  • Google Gemini (through direct API)
  • Anthropic (through Google Vertex AI)
  • Oracle Cloud Infrastructure (through Oracle)
To add the credentials for available LLMs, do these steps:
  1. Go to AI Agent Studio and open the Credentials tab.
  2. In the LLM subtab, add the API key.
  3. Add the details for the provider and save it.
    • Model: Choose the model you need from the available options.
    • API Key: Add the API key for the selected model.
    • Basic URL: Enter the endpoint URL that will receive requests for the LLM.
Your LLM now appears in the LLM tab, and you can select it when creating agents, nodes, or agent teams.
To request a new LLM model, do these steps:
  1. Create a service request through My Oracle Cloud Support.
  2. Add the details for the approver.
    • Supported provider name shown in AI Agent Studio.
    • Requested model and version.
    • Business use case.
    • Required timeline.
  3. Submit your request.

The time required to make a requested model available in AI Agent Studio depends on the complexity of the model integration. In most cases, requests are completed within two to four weeks. Each request is reviewed, and an estimated availability timeline is provided as part of the service request process. The model will appear in the Models list, where you can add LLMs.