Beth Lyons

Right.

And, and the thing that I read, the summary that I read said, if we, if we remember six months ago, the scores that they were getting were 5%.

Right.

And they, first of all, it's the second test because the open AI model, uh,

Did the first test, right?

So we had that.

That was actually last year during maybe, maybe during shipments.

So this is the second test that was created to be the harder test.

And these new results are over 50% when six months ago they were 5%.

Well, but the glass ceiling is like where we are right now is still a new frontier.

We're just starting to see the models that have been fully trained as multimodal, right?

Gemini 3 Pro is

had such a leap and also nano banana and I really think that's part of part of the reason for that is that it was given a multimodal diet to train on not a language model diet and then introduced to the concepts of the other modes and I think what just came out from China was also that but I feel like we're still in the very beginning of those and that we're going to see a lot

Like we may be under a glass ceiling, but I don't think we're going to be bored.

I don't think we're looking at the ceiling.

I think we're still like, wow, all this stuff in this room.

Let's figure it out.

Your co-hosts are here to help capture that as a permanent memory.

Yeah.

I also think, Brian, for their coming out with this as a push, we have to remember that OpenAI is headed by Sam Altman.