GPT 5.5 is here. It's all the rage. And Jaden, you know, I was just writing about this today or yesterday. You know, don't, I wouldn't, I would encourage people to keep an open mind when it comes to just like AI models and things like that, right? Because think about where we've been over the last week or two weeks. So Claude, the reigning champion, all of a sudden 4.7 comes out and people are like, eh. And then people are sort of saying like, oh, over the last couple of months. And then even Anthropic is sort of saying like, yeah, maybe it hasn't been that great over the last couple of months, even though everybody's been shouting about that. And everybody's been saying like OpenAI is dead in the water. And all of a sudden 5.5 rises up. People seem pretty enthusiastic about it. So, Jaden, when I'm looking at different models and how people are feeling and things like that, I just always encourage people not to get too comfortable with one model because even when one sort of like AI lab almost looks like dead in the water or something like that. By the way, I have news on Copilot, too, to that, speaking of which. True. Suddenly, suddenly they kind of come raging back. OpenAI seems to be coming raging back with 5.5. I know you are like playing around with it. What are you finding with 5.5? so there's a couple things that I think it can do and by the way I've been I'm probably like super annoying to all the frontier labs because as they fall behind I'm like very publicly like guys open AI is toast this is old news anthropic is where it's at and I mean to be fair if you always want like the absolute edge on a model and you want your model to be 10 15 percent better than what was then what else is on the market listen to my listen to my advice when I say that But if it like if the 10-15% improvement doesn't make a big difference, and you're locked into one model, like that's not going to kill you either. So anyways, that's my full disclosure. What 5.5 is really good at and that they're really trying to target here is I think a lot of these agentic multi flow coding type tools. In a lot of their demos, they're showing how you can, you know, someone said create a UFC tank shooter game. And we've kind of seen these like create like a little game thing before. And and basically, they're just getting better. They're just more advanced, just better at reasoning. It can just do things in less of an amount of time. So they have a whole bunch of really interesting demos. But I think basically what most people are impressed by is the fact that it's able to do agentic tasks. So what I've been using Anthropic, Opus 4.6, 4.7 with like co-work to do, it's able to do that kind of stuff. And I think OpenAI realizes that they have been slipping in kind of like users and in enterprise, probably specifically enterprise, agentic use case stuff. And this is their answer to that. I think it's a phenomenal model. I've been playing around with it. I'll show you some demos of what I got out of it in a second. But the one other thing that I wanted to mention with this is just because Anthropic has been doing so good with enterprise and with agentic, if you look at the, if you look at the, so there's basically these websites where you can buy, quote unquote, buy Anthropix shares or OpenAI shares off market. And typically how this works is it's kind of like, it's like employees that already work there and they have these shares that it's called secondaries. There's kind of these secondary markets. And sometimes it's like you'll buy a secondary share and maybe it doesn actually transfer right away But if there is like a event at the company then the share value increase falls through to you whatever there like a whole marketplace for these kind of secondary shares The reason I bringing it up is because OpenAI on the secondary market their company is valued at $850 billion and Anthropic is valued at $1 trillion. Is this the actual value of the company? Is this what it's actually worth? This isn't a commentary on that at But that is what people are willing to pay to get their hands on Anthropic versus OpenAI shares. And I think that is the red flag that or, you know, the code red that OpenAI was looking out for when they came out of the gate kicking with GPT 5.5. I mean, let me just say a word on that, too, because if you look at how Anthropic has grown, it's unbelievable. I mean, the ARR has been 10xing, 100xing. It's sort of like 1 billion to 10 billion to 100 billion. It's absolutely crazy. and even just anecdotally uh oh god jane some of our friends either over i can't remember it was maybe over at the neuron or maybe rowan over at uh the rundown or something these are all these some of these big newsletters i can't remember who ran it but they were sort of like running a thing of like hey what do you prefer and anthropic was twice as popular as chat gbt it's just a very strange place from where we were because you know open ai absolutely held this market completely now two things on this. First of all, we have to remember like people in the bubble versus people outside the bubble. There's still most people outside the bubble do not know really what Anthropic is. They're starting to hear about, they started to hear about Claude over the last couple of months, probably mostly because it got into the zeitgeist because of the, of the stuff that they were doing with the White House, where like the White House was saying, hey, this is like a, you know, supply chain risk and things like that. But also Claude was starting to, I'm sorry, Claude Code was starting to make it into the zeitgeist a little bit as well. But really, like if you think about it, like OpenAI still has 900 million weekly active users or something like that. And when people think about AI, they literally just think about ChatGPT. And this is where I think ChatGPT is starting to kind of like make a comeback. It just can't lose enterprise. Like that's the big thing because what you're hearing over and over again, even just anecdotally on X, Jaden, you've seen this too, just even inside some of these big companies, people want to use Claude. Claude, even though, you know, Gemini is incredibly powerful, but it just has felt like, Jane, over the last five months, it's been Gemini shipping like crazy, Claude shipping, like, or Anthropic shipping, like crazy. But now let's just take a breath and look at OpenAI over the last week. Well, I don't know when their, when their image model came out. I mean, but that was like pretty recent. That looks absolutely extraordinary. I'm seeing some people from OpenAI playing around with it. And 5.4, what was that four weeks ago or something like that? Like Like when we think about the pace of that, and also guys, when we think about this, I also just want to throw this out here because I want to see what Jayden was doing too. But like, you know, back in the day when GPT-4 kind of like held court for, I mean, it must have been 12 months or something insane, right? Like GPT-4 was just the absolute standard and people kept saying at some point, you know, we're going to have GPT-4 local on our phones and then you'll just be able to use that. But that hasn't panned out because people just want the highest model. People aren't even satisfied with Opus 4.5 anymore. They're like, I really want 4.6. I'm like, it's not a, the leap isn't huge right now with 4.7, but, but people just want the cutting edge. They don want sort of like the older model living locally So when I think about what open AI is doing it just means what I think it means anyway when you have this new model coming out is that the base level the worst level the free level of open AI is going to get better and better and better less hallucinations less everything else. So do I still have problems with open AI? Yeah, but tell me what you've been doing in your demos. Okay. So, I mean, something that I just ran with it because I like to test these models on kind of big tasks now. It's less of like write this, you know, rewrite this paragraph or check my email it's more i'm trying to get it to like do like a complicated task now which got to do research got to look at code and stuff so i mean right inside of chat gbt just having it on thinking mode with 5.5 i said do an seo site audit for ai chat daily.com it's a new vibe coding project i'm working on i'll probably talk about it next week or something you've been telling me offline yeah guys okay let me pause all that guys i just want it so i want you to blame jayden not me i right before we came i'm like jayden can we talk about this he's like uh I don't know. Not yet. Anyway, sorry, Jayden, go on. Okay. Basically, I'm just printing it up, so it'll be nice when people do see it. But what I did do is I wanted to do an SEO site audit. And I've actually been doing these things. I've been in marketing for a million years. That's what I've always done. And I've done SEO for so long. And so much of it, it feels like it's kind of like this black magic voodoo. And there's all these like crazy things. Claude is so freaking good at SEO and in a very programmatic way where it like restructures your whole site and all your pages and your internal linking and like all this crazy stuff. So I've actually been using it and I've been really impressed. I've been reading all of its docs whenever I ask it to do something, it will describe exactly everything it recommends. And you would have to spend so much money to get a company to do this, these kind of analysis for you, by the way, and implement them. And I have, by the way, I've spent thousands and thousands of for many years over all the projects I've ever worked on on this kind of stuff. Tens of thousands. Okay. So I wanted to see what, and I've been literally getting Claude to do SEO audits on this exact same site, multiple of them implementing everything. So I'm like, what would OpenAI give me after I've already done like five audits, right, on the site? And anyways, so I asked it for recommendations. I was really impressed scrolling through it because basically what it has to do is go to the site, scrape the site, check all the pages, check the site maps, check the code. There's a lot of steps that go into it. It gave me a very robust, and by the way, this feels so much less fluffy than GPT 5.4, where it's like, let me help you with that. Let me give you a no-fluff version of that. I'm like, don't tell me you're going to give me a no-fluff version. Just give it to me. Anyways, I think they fixed that. That issue has been patched. My wife and I were complaining about it all week. It's been patched with 5.5. The SEO audit that I actually went through and read flagged a bunch of great things that Claude didn't flag. So I actually think it came up with some new ideas that I'll further optimize. It basically gave me a very in-depth document that if I'm being 100% honest, I'm going to copy this full document, give it to Claude code where my tool is currently being built and have it implement everything for me. And I'm actually wondering if there's either two takeaways I'm getting from this exercise that I did using this takeaway number one might be ask the same question to multiple ai models by the way that i mean i wasn't trying to set myself up but plug for ai box.ai where you get 80 plus ai models in the same chat thread so you could ask them all the same question so that might be the answer um right because it seems like opening i caught some things that claude didn catch but answer number two which I think might also be a good point is anytime and this is maybe sort of annoying, but anytime a new model comes out, GPT 5.5, re-get it to do the same task you gave GPT 4, because I probably have asked GPT 5.4 to do SEO site audit before. Get the newer model to do it. And if the benchmarks are correct, like the software engineering benchmarks are correct, and it's like, look, it got 10%, 15% better. Maybe it's gonna give me 15% better advice, catch a couple of things it didn't catch before. And my website or my project just got 15% better as well. It kind of sounds annoying because it's like, dang, I gotta keep redoing the stuff I did before. But it genuinely gave me good ideas and good insights that I hadn't gotten before on this specific project. So I'm wondering, those are probably my two takeaways. Ask the same question to multiple models today, but anytime a new one comes out, if it's a really high value project that you're working on, it's probably worth reviewing. The final thing that I'll say on all of this that I did think was interesting is at the bottom of this question, if you're following along on YouTube and you can see the video, at the bottom of this whole question I have, it says, ask our best model GPT 5.5 Pro. It's tuned for hard tasks, science, data, business upgrade to unlock the best model. If you click upgrade to pro, you are hit with uh 100 a month they lowered it used to be 200 but i think open ai is i mean this is pretty aggressive they cut advertising that they were going to do right so i said we're not focused on ads instead they're just going to try to push you to the hundred dollar a month tier which was like claude does very aggressively so anyways it feels like we're in this push where it's like get the best models but you got to pay for them although in fairness claude i run out of tokens with claude all the time and or claude just goes down a lot it's been driving me a little nuts. I get why over capacity, but I actually took down my pro subscription to ChatGPT and this may get me back up to it. First of all, cause I didn't love paying a hundred, 200 bucks a month, a hundred bucks a month is actually easier to swallow. Uh, but it also, and guys, public service announcement on this, like when you're using these models, what Jaden said is, is really important because it's a step, it's a step that we often skip. I will be honest. I do too, but make sure that if you try like a new model, don't just give up on ChatGPT, be like, Hey, we released a new model, go back and try it. It doesn't really cost you much. Just like go back. Also make sure that you're clicking into the model and making sure that it's either sort of like on thinking or because right now the instant is 5.3, which is actually really good for the instant model to have 5.3. I don't think it was that until kind of like recently. But guys, most importantly, this is what I do all the time. I'm testing models against each other, which is why AIbox.ai, it's just the best thing that you can find out there. Because again, what I do, Jaden, is I have like my co-CEOs and it's like, it's been for the last couple of months, it's been Gemini and Claude. And so what do I do in AI box? I can actually just like compare these two things right back and forth. I can just sort of say like, Hey, what do you think of this? What do you think of this? It's literally like having a team. This is guys, this is not an expensive monthly thing. You can do this for Jane. What is it? The monthly, what's the monthly on AI box? We got you for $8.99. Yeah. $8.99. Guys, I just can't tell you how powerful this is, but it's not even just that. I mean, it's all the models. It's all the image generation, but you can also build workflows in there. It's insane that not every person on the planet has. I know you guys are going gangbusters, Jane. I know, I know, I know. But please, guys, if you haven't checked it out, make sure you check it out. Otherwise, I'm going to try to get stuff out of Jaden for maybe next week or something like that. He has some cool stuff to share. We will see you on the next episode.
