Deploy LLM model on new Foundry resource instead of AzureOpenAI

Hi Matt, Hi Pamela, Currently, the LLM is deployed as an AzureOpenAI resource. I wonder if I can just deploy it as a foundry resource. That way I could connect Azure AI Search to a foundry agent and have different tools connected. Do I just have to change the main.bicep file? Could I then be able to switch easily between deployed models that use the response API? (gpt 4.1, gpt 5.4, claude) I am not sure if this is something that a lot of people are interested in, but the new foundry experience offers a very nice user experience that people might prefer.