I see this so often. People try to show what GPT can or can’t do without explicitly stating the version. Since not many people have access to GPT-4 it’s usually

I see this so often. People try to show what GPT can or can’t do without explicitly stating the version. Since not many people have access to GPT-4 it’s usually safe to assume they didn’t use it. GPT-4 is a massive improvement over 3.5 when it come to any kind of non-trivial logic or inference task.