Cory

I agree.

I agree.

It is making significant progress.

Like the best, most efficient combination of those three?

For a million tokens, assuming five to one input output tokens ratio.

Is 3 Pro still the highest intelligence?

Depending on your benchmark of choice?

So they are neck and neck.

But if we're looking at that and compared to, you know, cost per API call or cost per token, yeah, I was just curious.

I wasn't sure.

You know, when those benchmarks first launch and –

uh i see and i don't feel like 5.2 has lower performing than 5.1 now that may be the case in some specific areas but in what most of what i do which is writing and creative endeavors and video stuff i get madly impressive results compared to what i did from 5.1 um

But I have seen that I have seen like, like, you know, there's always disagreement within the community and I love following those discussions because, you know, and maybe that just comes down to what's right for me what's right for you what's right for Tom and Tammy and whoever else I guess.

you know uh so often i think and and we talked we did it again today and talked a lot about software engineering without talking about other things people use ai for i there's this this tendency to lean really heavily into look how great it codes

and and i think we tend to forget that you know how well it does research how well it works in a spreadsheet how well it does this can really matter and and and as it gets better at one thing it seems that quite often the the adverse reaction is it gets less good at something else um at least that's that's been my experience and i talked with

well, I don't want to put him on blast, but I talked with a researcher a couple weeks ago while I was at Amazon, and he talked about that, that there's this give or take, give and take there that, you know, if the more you're optimizing for any one task, the less optimized you are for many, many others.

And I'm guessing we all look at every one of these through the eyes mostly of what

We do.

So there was a lot of optimism in there.

You know, one catch though, I think is something we talked about and we've talked about it.