GPT Sol 5.6 vs Claude Fable 5 For Marketing (Which Wins?)

But first, let's talk about this whole Anthropic and Blackstone bet on the next trillion-dollar AI business. So, you know, you've seen Anthropic, OpenAI, they've set up separate businesses and they've partnered up with like, I think they partnered up with like Ernst Young and these other like, what is it, Accenture or all these other consulting firms. But then now Goldman Sachs is involved, Blackstone, and they're helping fund this thing. So I think it's interesting because this group, I think it's called Ode or Ode. Which is conceived by Blackstone. They noticed the gap when it had roped in large consulting firms and small AI services to implement AI across portfolio companies. One of these boutiques, an engineering services startup, stood out and then they formed a joint venture. So guys, like, it's like, oh, why do people need the agency? Oh, can anybody just build this? Yes. But like I said earlier, more people are going to build stuff. Therefore, more people are going to be hired to maintain and optimize.

Dude, yeah. Did you see the OpenAI deal on the funding? I think they raised $4-point-something billion or whatever it was. It was something crazy ridiculous in which they're guaranteeing like 15% or 17% returns.

Wow.

Even OpenAI, right? A company, whether you say, oh, Anthropic's kicking their butt or not, to guarantee that much, right, in returns just shows you how hard it is to get money at really favorable deals when you want that kind of money and that business isn't proven yet. Like, people just expect, and I don't blame them.

So I'm going to show this first. So Saul 5.6, in my opinion, has been kicking Fable 5's butt in some aspects of marketing. So I did, I did, I ran some tests over here. And so basically for a website build where they were building the same website, one took about 34 minutes, which is Saul, and then Fable 5 took 3+ hours and it kept getting stuck and it kept asking me questions back and forth. So imagine that when you're using the Frontier, uh, GPT's Frontier model 5.6, um, it could— it's just the employee that doesn't have to do all this back and forth and it just gets the job done. The other thing is it does a really good job with thumbnails, right? So Saul, it shipped everything same day when I just stepped away from the computer. Fable 5 failed to download and I kept doing all this back and forth stuff again. Um, I also had Saul do clips for marketing school. So Neil, our marketing school from last week, I think it was like 1 hour, it was like 80 minutes, right? So I had Saul, I was like, hey, find— I want you to find the spikes from the content and then put a hook in the beginning of it and I want you to chop it up for short form. It literally, it was one simple prompt and it actually rendered 10 of them and it actually looked good. Fable 5 got stuck in a loop, right? So, and some of this other stuff over here, I had Saul do an Amazon growth analysis and Saul took 9 minutes over here and it ranked the best. Fable 5 was just as strong, but it took way longer. And here's the crazy thing, Neil. When I had the Amazon analysis on what, as an investor, what area would I bet on if I was investing in Amazon or if I was running Amazon, Sol said it would bet on Amazon's new fulfillment services, like supply chain services, whereas Fable said it would bet on same-day grocery. Which one do you think is more appealing? Same-day grocery or The supply chain services where you can basically leverage Amazon's logistics network to ship whatever you want.

The logistics is boring. Usually the boring stuff makes more money.

So I'm with you, right? And so Saul did the analysis faster and it came up with a better strategy. And so what I saw with my limited experimentation here was that GPT Saul did a better job when it came to marketing. And I just want to show you some thumbnails before we move to yours. So look at these thumbnails, Neil.

Like, This is— hold on, what's this thing?

Okay, you see this?

The real work test. All right, SOL versus Fable 5. All right.

Yeah. And then this one over, like these over here, like it actually took our— it took my transcripts from a longer form video and it made these thumbnails. Like it's like saying, oh, SOL 5.6 is faster and cheaper. Okay. But then SOL is like, you need to request all these permissions. So this is easy. I'm just chatting with the thumbnail designer back and forth, whereas Fable 5 can't even do this at a reliable level right now. And this is a Fable 5 logo. Notice that I changed the Claude logo. I said, I want the Fable 5 logo. Boom, you got this. So anyway, all that to say, your mileage may vary. I think you should test both. But, you know, these things are just getting better and better.

Yeah, they are getting better and better. People say OpenAI's version, their new stuff is better than Anthropic's. At least that's what I'm hearing from a lot of engineers. But I have no clue.

So I use— I secretly use my Codex, or it's not called Codex anymore, ChatGPT work, but we pay for Claude Teams.

So yeah. Which one do you like better?

I like ChatGPT more right now.

Would you make your team switch?

No, because it's too much of a pain in the butt to do that logistically. So I'm not going to do it. And but I would just say that anybody listening to this right now, I would suggest that you don't lock yourself into a long-term contract with Claude or OpenAI because they're just constantly leapfrogging each other. And then, yes, yeah, so—

You nailed it. And that's the way I take it. I'm like, it may be better now, but who knows if it's better or worse in a few months from now.

Plus the American open weights that are coming out now, like Meta's stepping into the game, you know, NVIDIA stepping into the game, uh, Thinking Machine just came out with theirs. And so I think we're all going to be using a lot of open models and we're going to be using a lot of model routing as well.

Yep.

All right, so what did you find?

Uh, so they raised $4— it was either $4 or $4.5 billion. I asked Google the version 3 times and it came up with the answer faster than ChatGPT because I can just modify it really quickly. Uh, and they ended up raising, uh, $4 or $4.5 billion with the guaranteed 17+% return. It was either 17 or 17.5. It was a massive return that they're promising.

Wow, that's crazy. And the other thing too is like, um, all the deal— like, I think Sam Altman, uh, offered, um, the United States 5% of OpenAI or something, just maybe get that, that, that edge, um, over their competitors. And then obviously you have like, um, Anthropic doesn't want these open models to, to, to, to come out and gain more traction because that affects their business model. The other thing that I think is really interesting, Neil, is just from a business and pricing standpoint. Is when you look at— so Fable 5, by the way, as of July 12th, they were supposed to just move it all over to API. The challenge is ChatGPT just came out with SOL 5.6, and then, you know, they're just going to build into the model and they're not going to charge you API usage for it, at least from what I see right now. So now they're making Claude dance. It's like, oh, Claude's like, oh yeah, we reset your usage. It'll be till the 19th, July 19th, before we move it to usage. Now they might not even be able to do that. And so you see, like we've talked about this, where Gemini might be squeezing everyone by just bringing the pricing down and making it free for everyone or cheaper. And then you have OpenAI where it's like, oh, we're going to make Fable 5, we're going to make Claude Dats, and they're doing that right now. And then now like Anthropic is like, oh crap, maybe we can't move it to usage only, which was what their plan was.

So. Dude, I just can't believe how much money these guys all raised. And I'm curious what ends up happening in the long run.

I think, you know what, uh, do you— have you heard any of Gavin Baker's stuff? The investor?

No.

No, super smart. Like super smart guy when it comes to AI chips and, and like bottlenecks. Like everyone's doing this AI bottleneck, uh, trade right now. We're talking about it.

Um, so his whole thing is this bottleneck trade where they're just buying the stocks for the companies, uh, like the memory stocks, the ones that—

Exactly. Anything that's where it's constrained right now in the supply chain. Um, could be memory right now, could be cooling, for example, it could be other things. Um, but He said this, during the dot-com bust or dot-com boom before the bust, everyone was talking about you got to build out more fiber, more fiber for the internet, right? And then you know why they call it dark fiber? It's because 97% of that buildout was unused.

Wow.

But today you don't have any dark GPUs. All the GPUs are being used and more and more and more are being used. So I don't think the need for intelligence is ever going to come down because, Neil, keep in mind when you and I got broadband, what happened? It wasn't that we just got to use the internet faster. We got streaming, right? We got to do a lot of other things. We got voice communication, all these other things. So I think it's going to spawn a lot more. And I was listening to Howard Marks' podcast today and he's like, the interviewers were like, so when do you know when to bet hard, right? And he's like, well, you never really know, but if you know that investing will— not investing is going to hurt more, then you better make sure that you're investing. Otherwise it's going to be a bad situation. Especially because he has a firm.

So I think these all things kind of— Oh, he doesn't have a firm anymore. He hasn't had a firm for a while.

Howard Marks?

Yeah.

You're talking about Oaktree?

Yeah.

Yeah. So this was back in the day. He was just saying back in the day when— yeah. When, when, but because he was scared, right? But he's like, well, if we don't do this, then we're not doing our job. So we need to do it no matter what.

So. Yeah. Uh, let's see. He didn't sell. Okay. Well, he sold it in 2 parts. The first in '62%. The final sale was actually recently, October 2025. Well, I know he's balling because he owns a home by me and it's a nice home. And then I believe it was one of his kids that wanted a home, so he bought the home next door and expanded and just hooked his kid up. He is a billionaire, so he could do whatever he wants.

Yeah. Sounds like something you'll do at some point too. Okay. So, okay, we talked about Anthropic's bet, I think. So just to be clear with everyone, Neil and I are saying that services aren't going away anytime soon, which is interesting. Neil, I was telling you at one of the conferences, one guy, you know, wanted to sell his company to potentially you.

So— Oh, the—

Don't reveal.

Don't reveal.

Huh?

The Spain guy?

Not the Spain guy. No, I gave you another name on Tuesday. It was an SEO company.

Oh, yeah, yeah, yeah. I know what you're talking about. I thought you were talking about the Spain guy that we met at A conference a few years ago.

No, that wasn't a fit for you. Yeah.

It wasn't a fit because it wasn't growing. And then I talked to him on what he would take and he said, yeah. And then he told my team he wanted like 8 times profit for a declining business. I was like, yeah, this ain't going to happen.

So Neil, let's workshop this company. I mean, you know, we'll just give, we'll just give high-level numbers. Let's say this company is doing, let's say $4 million in EBITDA. Okay. And let's say the profits are, sorry, the revenue is $15 million or so. So this company considers themselves AI native, right? And I think that's something that you guys would like to have. And I think this founder's a very smart person. So what's your take on this one? Because you had talked to this company before. All right. So I wanted to take a moment to tell you about my podcast co-host Neil's agency called NP Digital, and they work with a whole host of global companies or global organizations. Also, Neil has SEO tools such as Ubersuggest and AnswerThePublic. All you have to do is go to npdigital.com to learn more, and we'll see you on the other side.

Um, I haven't talked to them in a long time, so it's a good company. Uh, I think they'll be able to sell if they can maintain the growth. They can probably get maybe close to $30 million including earnout. They won't get most of the money upfront, but they can get a good chunk upfront. The problem with AI-native consulting companies is, and this one was an AI-native consulting company. I think a lot of the revenue is non-AI. They are just pushing on the AI narrative more now. You have to look at what portion of their revenue is from these new services and how sticky is it. But I think a lot of these players will have a ton of competition, and then if their margins are high, call it 25-30%, I think there's going to be margin compression over time. The other thing that you have to look out is if the company is relying on one person, we call that key man risk. Sometimes it could be an influencer, sometimes it could be a well-connected CEO who ran a publicly traded company, has all the relationships, the list goes on and on. Sometimes it could be someone who published a book and they're a New York Times bestselling author and people read the book and that's what drives all the leads. But if there's Key risks like that, you'll start seeing the deal starting to get dinged. And I do believe over time when people have higher margins, it's because they haven't built out their management team to reduce the risk.

Yeah.

And we constantly see that when agencies running like 30, 40% profit, like I would say 4 on 15 is high profit, and the chances are is they don't have as well established of a management team. That would be my guess on You know, 26, 27% margin type of business and services.

And just so everyone knows, I mean, those, the percentage, the numbers I'm giving are, you know, modified a little bit so it doesn't reveal anybody. What's interesting, Neil, was that earlier this week I was talking to an AI consulting firm here in New York. So let's just say they're doing about $25, $30 million a year right now, and they are getting VCs hitting them up everywhere, and they don't want to raise money, right? And then they're like, okay, maybe they want to take on debt because one guy sitting at the table that I was talking to, he actually funds a lot of businesses, right? He actually— it's a business funding business, business loans business, right? So again, going back to the AI consulting firm, they're getting people hitting them up left and right. And then they're just like, wow, like, do we just take on more debt? But that's scary. I'm like, no matter what you do, you need to take on more money right now. My opinion is whether you raise venture or you take on debt, my opinion is If you don't, the problem is I just look at all these logos that you're going to pick up right now as expansion opportunities. And if someone comes in and vacuums them all up or vacuums a lot of them up, it's not to say that there isn't a lot of business everywhere. There certainly is. But I think this is a nice land grab opportunity in the next year or 2. I don't look at it as land grab as like Meta or like Facebook, like that type of land grab. But I think it's still very much a land grab.

I think they should take a totally different approach. They shouldn't take the debt if, because if the guy's afraid of debt and the risk and he's already built something amazing, I'm guessing this is his first big company that he's built.

First big company, but he's raised venture for his last few companies and none of them worked out. Yeah.

So what I would do if this is the first big company that he built, I would go and not take on more debt. I would not go and raise venture capital. I would go sell a minority chunk to private equity. He'll take chips off the table enough where he never has to work again. They'll put in money and debt into the company so that way he can grow and build something even bigger. If it doesn't work out, he's made enough.

So you're saying, you would say he should go build a platform with a PE firm?

Not necessarily build a platform, he would be the platform and they would tuck in other companies to just make him larger. But I would go do it with private equity. That's his best bet because if he sold 49% or 40% in a minority and they put in more capital to grow faster, and he's going to get diluted as they buy more companies, especially if the PE has to put in more cash, but he could take enough money off the table where he's okay. And if he really thinks it's a big opportunity and the amount of money he can get right now is enough where he's happy with it, then he should sell majority and take 70% off the table. And then if he takes 70% off the table, They'll go buy more. And typically if it's something hot and they do well, whatever he made on the front end, even if on the back end he only owns 30% and that gets diluted down, the back end usually will be a similar size to the front end. So then he can make double the amount of money. And if the front end was big enough that he never has to worry again and his kids don't have to worry again, he's really good if the back end succeeds. Yep.

Or bigger, which could very well be the case.

But yeah, I also think a lot of these companies will go back down to normal valuations, just like, uh, I believe it was Scrunch that sold for $200-something million or something crazy like that on $20 million in revenue. Scrunch, the Profound competitor, right? Oh, it was something like that that got reported.

Wait, how much was it?

Like, okay, if you look at Profound, Profound raised well over $100 million, a billion-dollar valuation. Let's say they do $50 million in revenue. Do you think that valuation's going to keep climbing? I'm not saying they can't sell for a lot of money, but I'm saying 5, 10 years from now, do you think that valuation's going to be the same or it's going to come down to— it's going to be normalized?

I think it's going to come down to earth because you and me being SEOs, I don't think it's something that— to me, it's like the new SEO and I'm like, how does this— how is this adding up? But we'll see. But they raised over $155 million.

Totally agree with you. So I look at a lot of these things, including AI consulting or implementation businesses that are really heavily valued right now, but in 5 years there'll be a ton of 'em and the valuations will drop down and they'll be normalized.

Yeah. And so what Neil's really saying here is that strike while the iron is hot right now. Take some chips off the table while the valuation is hot.

Because he could be like, we're growing really fast, but private equity's probably gonna value his company double what it really should be valued. So then you just take 70% off the table.

Anything that grows really fast, what do you think is going to happen? Of course it invites competition and that means what goes up must come down.

Yes. And then you have people competing on price and, you know, giving away some of the things for free. We saw even with SEO, dude, a lot of the agencies back in the day would say, oh, SEO is hot on you. No problem. If you just do paid media through us, we'll do the SEO for free. Of course they didn't put enough resources into it because they're giving away something for free, so they barely did much. But still people are like, oh, okay, sounds good. I'm getting this new shiny object for free. Now granted, SEO is much older now, but people start doing creative things to get the business. Yep.

All right, so we got another 5 minutes left here. What do you got on your side?

All right, so, you know, right now we've seen a huge change in creative. Okay? And when I say creative, when you're running marketing campaigns, you're generating Facebook ads, you're generating Google ads. It could be video ads, CTV, so like ads on Netflix. And I remember a few years ago everyone's like, you don't need creative shops anymore. You can just have AI do everything. Now I'm giving creative as an example because that could easily be disrupted by AI. And it has in many cases because you can use humans for post-production. I mean, you can use AI for post-production and a lot of these other things to create more We're seeing a whole new shift in marketing right now. A year ago, companies were putting a lot of pressure on marketers and agencies. If I actually look to now, customers and marketers are actually getting less pressure about AI than before. We also see a lot of big corporations slowly remove or starting to cut in certain areas because they were getting pressure from finance. at least in marketing on headcount, why can't you do more with AI? To now they're rehiring because they found that AI is creating a lot of slop and they still need the humans involved in the process. But what I'm getting at is whether it's creative or whether it's just general marketing tasks, yes, AI has affected marketing, but we're seeing a lot less pressure from customers and just companies alike that everything needs to be AI forward and you need to do more for the same money and you're not doing enough and why are we using humans? A lot of them are actually reverting, not to back where it was, but somewhere in between that is more realistic. And I would have to say the pendulum is swinging more towards the human side than the AI side, from what we're seeing from a pressure perspective from companies.

That's interesting, because earlier when we started this podcast, you mentioned kind of the do more expectation. We called out the survey as well, right? The expectation is to do more from founders. But I'm like thinking, I'm like, okay, well, Wouldn't that be the expectation too from clients too, to do more fees? But maybe not so much like because you're the client is not necessarily the founder all the time. So I don't know. That's just a thought that I had as you were talking about it. Yep.

But realistically, majority of the calls we have, people aren't pushing as hard as they were a year ago. And the reason being is that they weren't getting the results from all this AI investment. I'm not saying that won't change, but they were pushing really hard, expecting people to be super AI forward. And when they look at the same budget that they spent with this new model, they're not really seeing any more growth. And when people started testing out some of the older models and the older ways, they really weren't seeing much of a difference either because a lot of these big corporations have like 20, 30, 50 divisions so they can run experiments, right? And it's the same agencies that they're paying that use technology versus that don't, or they select a few different ones. And what we're seeing is people are using AI in marketing. People just have to accept that. And people are accepting— expecting good results and a little bit more. But the crazy thing before was like, you should be able to double, you should be able to do triple. When they're pushing that narrative, the quality just sucked.

Yeah, the way I see it before we wrap up here is Services are here to stay. AI is here to stay. But what is also here to stay is you're always going to need, for the most part, human in the loop to quality check, to optimize, to maintain things. That doesn't change. If people are going to build more stuff, you're going to need more maintainers and optimizers and strategists. And so that's why I think Neil and I'm speaking for Neil as well. We see that more humans, we're going to need to hire more humans to do work, and a lot of people are going to be reskilled over the next call it the next 10+ years or so. So I think it's a net positive, and I think people are still getting used to it. And then the good news is, to Neal's point, expectations have dropped a little bit because this stuff has started to normalize. Anything else, Neal?

That's it. Thank you guys for listening in. We'll see you tomorrow.