Caroline Hyde

How are you competing against these vast generative AI companies?

I think we've managed to outcompete the largest research labs by being very focused.

I think it's the era of both research and efficiency.

And if you're able to maintain the focus as a team, you're able to deliver kind of groundbreaking results.

And we're proving this.

This is the first time I think we've, again, anyone has kind of topped the leaderboards not being a large, well-funded research lab.

And I think part of it is really the team.

And it's really also the vision.

We've been working on this for almost seven years.

We started working on video models when there weren't even the other boards to start with.

And I think eventually you built some sort of intuition and really good momentum as to how to improve these models over time.

And look, this is still the worst the models will ever be.

And so we have a bunch of more releases coming up that I think will further improve both pre-training and post-training.

And so we're very excited for that.

Chris Bell, you're not that small.

You raised $300 million in April at a foremost $4 billion or $3.3 billion valuation.

I do think there's some value in you explaining what was different this time around in the training of Gen 4.5 and the data set.

What is it that you did differently and that has allowed you to release such a competitive model?

I think there's a lot of different things.

On the one hand, pre-training has been one of our focus for a long time, making sure that both the algorithmic improvements are there, but also the way we caption, we structure data, we build the models themselves and test them.