Artificial intelligence can be classified in three levels: artificial narrow intelligence (ANI), artificial general intelligence (AGI) and artificial super intelligence (ASI). While the models of the current A I. graze can be debated to be real artificial intelligences the GPT style language models can be classified in similar manner. They might not be real intelligences, but they can mimic the human knowledge in convincing extend.
There are many smaller, more focused models that fall into the narrow category. They are specialized in single domain and can produce meaningful answers related to that knowledge, but can't really do much for more generic topics. As the more capable models still require substantial amount of computing power these smaller models are great way to bring the more limited knowledge to less powerful computers.
Most of the current state-of-the-art models fall into the generic category. They can discuss about almost any topic but the power requirements keep them out from anything but centralized cloud services.
While the advances in the field have been fast, there isn't yet a model available that could surpass human "intelligence" to be categorized as super.