Like or not, openai isn't having the same compute strain, meaning this was predictable.