Max output for older GPT-4 models is the same for input and output tokens. Only the newer API models are limited to 4096 tokens.

→ View original post on X — @officiallogank