OpenAI is not doing us any favours by using confusing naming schemes for their models and obscuring which models people are actually working with.
If I didn’t know any better, I’d say OpenAI doesn’t want us doing these tests accurately and is trying to hide something.