I literally told you my use case and it failed on that and it failed on similar problem 1-2 months ago when I was using 3.5 version, for my class of problems nothing changes, it fails the same way. I think you have your eyes shut and not reading what people write. I'm not talking about easy CRUD problems that you can find thousands of solutions online, ChatGPT is doing ok on these kind of tasks and it solved a lot of them for me too.