Mark Zuckerberg

But every new major model that comes out now is multimodal, right?

It's image, it's voice, and theirs isn't.

And now the question is, why is that the case?

I don't think it's because they're not capable of doing it.

I think that they basically had to spend their calories on doing these infrastructure optimizations to overcome the fact that there were these export controls.

But when you compare like Llama 4,

With DeepSeq, I mean, our reasoning model isn't out yet.

So I think that the kind of R1 comparison isn't clear yet.

But we're basically like,

Effectively same ballpark on all the tech stuff is what deep seek is doing but with a smaller model So it's it's much more kind of efficient per The kind of cost per intelligence is lower with what we're doing for llama on text and then all the multimodal stuff We're effectively leading at and it just doesn't even exist in their stuff.

So So I think that the llama for models when you compare them to what they're doing are

are good and I think generally people are gonna prefer to use the Llama 4 models.

But I think that there is this interesting contour where like it's clearly a good team that's doing stuff over there.

And I think you're right to ask about the accessibility of power, the accessibility of compute and chips and things like that.

Because I think the kind of work that you're seeing the different labs do and play out, I think is somewhat downstream of that.

I mean, look, we've basically pioneered the open source LLM thing, so I don't consider the license to be onerous.

I kind of think that when we were starting to push on open source, it was this big debate in the industry of like,

is this even a reasonable thing to do?

Can you do something that is safe and trustworthy with open source?

Will open source ever be able to be competitive enough that anyone will even care?