From a full day of testing with the new gpt-4-turbo, a few things stand out: It's much less lazy and more willing to output large chunks of complete code. It seems better at reasoning. Plus, function calling with vision is going to be a huge unlock.

→ View original post on X — @skirano