Basically, in order to stay relevant, Open AI and others are just going to do more meta trainings of smaller LLMs over LLMS, so you are going to get some specia

Basically, in order to stay relevant, Open AI and others are just going to do more meta trainings of smaller LLMs over LLMS, so you are going to get some specialized LLMS that do good for certain things, but you aren't going to see any massive improvement in functionality until someone fundamentally rethinks architecture down to the core.