[mcielecki]

Well, it's more than that, with o1 they moved to an architecture that hides its thinking process, uses an uncensored model AND hampers the ability for the user to control the outputs.

At least in the past we were able to see all the actual thoughts and chain of thought that the ai was going through.

EDIT: Furthermore they made it explicit that, even as the model is developed, they are not going to show the actual COT reasoning, even via the API.