Open source models seem to be closing up quick and the government is worried

Is there real possibility of open source catching up to frontier models? Months for me, that's still big, particularly in those government use cases we talked about. That's still more than enough time for an Anthropic or OpenAI to become a multi-trillion dollar company. What are we - how fast, or how, how smooth do you envision that gap becoming in terms of the the number of months between a frontier model releasing its latest version and an open source version being able to replicate similar capability? Yeah, at least a month, definitely. At least a month, maybe a month to two months. I don't really see it getting lower than that. And I think what's, what they're going to try to start doing, and it seems like what they're doing now is, they're going to be limiting people who has, who have access to those frontier models. They're going to say, 'Only you, you, you has access.' And so they can very well monitor to make sure that they're not able to catch up that quickly. Um. So, I guess in that case, probably the frontier models would only go to like the government, and like one of the top people, and later on they would then allow it to distill down. So the US government can probably always stay ahead, um, ahead of the curve. All right man. It's been a good one. I'm, I'm, I'm, I'm highly curious.