> Not my understanding of how the LLMs currently work. There’s a cutoff date and no new info after that date is included in training.

You are almost certainly correct in terms of how the public models work. I guess it's possible that the labs have some tools connected for more recent stuff like this, but it seems unlikely.