Carl

That, of course, is not us.

This is who the New York Times or Gemini Pro Nanobanana feels would be the good, the proper...

Yeah, the proper co-host, that's sort of funny.

But anyway, I am interested in how people are experiencing Nano Banana or these new Gemini 3 Pro models.

It does seem that this has been one that is helpful.

Ooh, Andy's here.

Awesome.

Let me shift our visuals.

It does seem that Nano Banana Pro, they've been talking about Gemini 3 and models that have the ability to, there we go.

Now we're in slightly better.

Hi, Andy, how are you doing?

Oh, no, I'm terribly sorry.

So we had our first single person daily show for like.

Yeah, that's tricky.

So anyway, I believe Gemini 3 regular was stated to be the first model that had been trained multimedia from the beginning, which means that it wasn't an LLM that then was given the ability to understand

images, it was given everything up front and then was able to create itself.

But it does seem that the pro version is much more that.

So I don't know if the 3.0 wasn't really that, it was still an LLM,

that was then just given the ability to understand images, or whether Gemini 3 Pro is actually the full version of what happens when a model is natively multimedia from the beginning, or multimodal.

So the image, so Nanobanana Pro is actually just make an image with the thinking version of Gemini 3.