To keep this experiment as fair as possible, I decided to stick with the default AI model in each coding tool. That meant ...