I am building a copilot virtual agent that basically asks a user a lot of questions, concatenates everything into a query, then uses automate to call the ai and respond. The query works find in my ai playground. It provides what looks like chat gpt-4 responds with. Very nice. But when I use automate action "Create text with GPT using a prompt" I get what looks to be an old gpt3 like response. Is there a better action to use? Or, where can I check to see that action is using the correct model? Thx!