I think open models trail frontier by maybe 6-12 months right now. But today's frontier already solves most practical problems. So give it a year — open models hit today's level, and that's good enough for most use cases. At that point the competition stops being about who trains the best model and starts being about who builds the best applications on top of models everyone has access to.
I think open models trail frontier by maybe 6-12 months right now. But today's frontier already solves most practical problems. So give it a year — open models hit today's level, and that's good enough for most use cases. At that point the competition stops being about who trains the best model and starts being about who builds the best applications on top of models everyone has access to.
Time to IPO before the rest of market catches on.