ChatGPT is surprisingly bad at knowing what model it’s using — it either confidently gives an incorrect version or at best it’s vague (“based on the GPT-4 architecture”). No current model seems to ever know its exact name.

→ View original post on X — @goodside