Does ChatGPT really know which model I use in api?

I used gpt-4-0613 model in my application and asked: Are you based on gpt-3 or gpt-4 model? The answer is always gpt-3, same output in playground. Why’s that?

Welcome to the forum,

GPT-4 was created after the training data cut-off date of 2021, so to it, GPT-4 does not exist.

1 Like

I think GPT is based on GPT-3.5 architecture, not GPT-4. The response you receive in the playground might be a default response indicating the GPT-3.5 model:) But yup, GTP’s knowledge is based on information available up until September 2021, so can’t have information about more recent versions.