00:00 Speaker A

Okay. So Anthropic.

00:01 Speaker A

Uh yesterday, there was a new Claude model released.

00:04 Speaker B

Yes.

00:04 Speaker A

4.7, Opus 4.7.

00:06 Speaker A

Um

00:07 Speaker B

I like how they give them each a name and a number.

00:10 Speaker A

And a number. That’s right. That’s right.

00:12 Speaker A

Miles 1.0, 2.0, 3.0. Um

00:15 Speaker A

I think a couple things. One, it’s like, oh, I guess they have a new model after they didn’t quite release the Mythos model last week.

00:19 Speaker B

This is like the watered down Mythos, the safe Mythos.

00:22 Speaker A

Okay. And it seems like if you go on X and you look at people talking about Opus 4.7, it seems like it was kind of tossed off. People are not super satisfied with this new model upgrade. Um, and I think that what’s interesting is the way the AI vibes have really shifted in the last couple of weeks. It was super pro-anthropic, really in the wake of the DOD dust up. Uh, Open AI kind of had, um, there was the whole, I guess that was right when like Fiji Simo stepped down and they did the the TBPN deal. It was like, what’s going on over it search on Sora, what’s going on at Open AI.

00:54 Speaker A

And now, it feels like what’s going on in Anthropic? Now they’re releasing models that don’t feel fully baked. A lot of folks are complaining they’re getting throttled on their token usage with Claude on a daily basis. They don’t have enough compute. The company has said they don’t have enough compute. Jensen Wong is coming out earlier this week saying, they don’t have enough compute. You know, sorry, we couldn’t make a deal with them. Like, it it does feel like this was not a great week for Anthropic and it is interesting how tenuous that vibe winner within the AI wars, how quickly that kind of feels like it goes back and forth.

01:19 Speaker B

Yeah, I mean it has flipped back and forth a bunch of times already. I haven’t picked up on that vibe quite as much as you have. Maybe it has to do with, well, I was also in DC in World Bank mode for a couple days this week and that was not

01:29 Speaker A

No AI vibe down there.

01:30 Speaker B

I mean, Mythos was a topic of conversation, but you know, that that wasn’t necessarily an anti-anthropic vibe. Um, so I don’t know that I’ve picked up on as much. My Claude defaults to the previous model, so I haven’t had the and I guess I don’t use it enough to have been throttled at this point. Um, so I haven’t picked up on that quite as much. There was I did notice, you know, Ramp, our the folks over at Ramp who monitor business usage of Open AI versus Anthropic and the other models. Um, Eric Karazanian over there, who’s their lead economist is estimating that Anthropic, um, its business, its enterprise, um, business is going to overtake that of Open AI soon. Yeah.

02:08 Speaker B

So like that’s still sort of implying in the background that Anthropic is still doing well, but

02:13 Speaker B

if they keep getting throttled, then maybe maybe that’s going to change. I don’t know.

02:14 Speaker A

Well, to go back to the

02:15 Speaker B

Maybe they’re preserving that, maybe they’re preserving that capacity for their enterprise customers, one would think.

02:21 Speaker A

Well, last thing on this because I know we got to go. But on the World Bank point of it. Yes. Not to, um, that’s a very specific crowd at the World Bank kinds of meetings. Sure.

02:29 Speaker A

It is not a uh an AI engineering crowd, let’s say, right? That’s not really they’re like a technocratic crowd. Yes.

02:35 Speaker A

And I think that what is also interesting in this anthropic like where what are we doing here moment, it’s like, okay, you have Mythos, which I guess kind of exists, but has been released in a limited fashion. You have bank CEOs being called in to have a meeting about something that like kind of doesn’t exist fully yet. You have technocrats at the World Bank talking about something that kind of doesn’t exist yet, and it feels like it’s so nakedly like this, you won’t believe what we have. And are we 100% sure that it actually is like, I can’t believe this. It it’s like every model has been amazing because every AI advance is amazing. And this framing that Mythos is like so amazing and uniquely dangerous in a way we haven’t seen after they release a new model, which people are like, this kind of looks like slop,

03:17 Speaker A

it feels like there’s just a sequencing here that speaks to this like, I’m not sure about, you know, the the why the vibe shifted. Like that feels to me why it shifted.

03:24 Speaker B

I don’t know. I mean, I think it’s entirely possible that they had to tweak Mythos and then come out with this new Opus as a watered down version and that’s why it’s less than satisfying to people. I don’t know.

03:33 Speaker A

Right, rather than it’s actually not what you say it is, but you release this other thing. So the new like you release something that’s not quite as good as you wanted to be because then the next thing will seem like a bigger leap than it would have otherwise.

03:40 Speaker B

Right. Maybe?

03:42 Speaker A

Maybe that’s my more negative way to read it.