Sometimes I'd like to switch models during a prompt chain/queue. For example, I may want to create a prompt to run the first part of the prompt on GPT-4o and then apply the results to the second part of the prompt and run that on Gemini 2.0.