NEW GPT 5.4 Update is HERE!

GPT 5. 4 is here. Open eye literally just dropped this today and we're going to be testing it out live so you don't have to figure it out yourself. This isn't a small update. We're talking about a brand new model that outperforms the average office worker 83% of the time. Can now control your computer. It plugs directly into Excel and Google Sheets. It uses nearly half the tokens, which means it's faster and cheaper. There's also um a new version called GPT 5. 4 Thinking. basically a supercharge mode where it slows down and goes deep on the hardest problems. And for the first time ever, it supports a 1 million token context window. That means it can read an entire book, an entire code base, an entire business, and actually remember all of it. I'm going to walk you through everything that's new, what actually matters, what's just hype, and whether it's worth switching from whatever you're using right now. If you use AI for literally anything, work, research, building stuff, you need to see it. Let's get into this and we're going to get started. So, you can see the announcement right here. GPT 5. 4

Thinking and GPT 5. 4 Pro rolling out in chat GPT 5. 4 is now available in the API and the codecs and GPT 5. 4 brings uh advances in reasoning, coding, and agentic workflows into one Frontier model. You can see the benchmarks right here. I don't know if anyone cares about benchmarks at this point to be honest. But if you want to get started with using this, you can go over to chat GBT and if you've got it and it's already rolled out to you, then you just select 5. 4 in the top, right? So you got pro and you got thinking. Thinking is like, you know, better for longer answers. Pro is like research grade intelligence. The only issues with pro is like usually it takes a long time to respond. So just something to be aware of right there. Now, if we have a look on OpenAI's website, we'll take a look and we'll see what announcement they've made on this. Here we go. So, GPT 5. 4. It literally it just dropped a few hours ago. Um, so it's supposed to be their most efficient and capable Frontier model. Honestly, like a lot of the models in the 5 series have not been that good. So, it'll be interesting to see how this performs. Finally, GPT4 5. 4 4 is the most efficient token reasoning model benchmarks here. Good for knowledge work. Let's

just try out see how it performs. All right. So, we're going to go inside here and I'm going to switch. You can actually change the thinking effort as well for pro. So, you can change between standard and extended. I'm going to switch over to the thinking mode because pro will take absolutely ages. I'm going to try and build this. Now, let's just check if we can get canvas on this as well. canvas will allow us to preview what we build, which you can, which is great. So, I'm going to start off by trying to build a hyper dopamine pong game, and we'll see how that performs. It's pretty fast when it comes out, actually. Not too bad when it comes to getting the answers quickly. What we can actually do is we can go over to Claude Sonet 4. 6. And, you know, we'll change it to Opus, uh, Claude Opus 4. 6, and we'll run exactly the same prompt. and just see which one gives us the best output, how they perform side by side and that sort of thing. So, we've got Chat GPT over here and we have Claude over here and they're both running the same prompt. We'll see how they perform. Damovicious says they just can't keep up with Anthropic. I would agree. Like, I mean, Anthropic is just it's too good, isn't it? Actually, I think they hit number one in the um app store charts recently. Anthropic, they went from like 131 in January to number one in February. So they're absolutely crushing it. I don't know if anyone saw the news, but actually 1. 5 million people quit chat in the space of like 48 hours recently. Pretty crazy stuff.

Now the other thing to note here as well is like it does pretty well on computer use too. So you can see an example here like it's operating someone's emails and then over here it's working on a bulk data entry as well. So it can like fill out forms and that sort of thing. So I think one of the improvements is that it's designed for computer use, right? It's designed to be much better at computer use as well. Let's see what we're up to. All right, I think we've got the outputs here. Yep. So, we've got the finished version of this pong game from ChachiPT. Claude is still building out, so we'll come back to that in a minute. Let's have a look and see how this performs. So, it's winning on speed when it comes to cord, but let's see if it's winning on outputs. So, this is the game itself that's been built. It's not bad. It's pretty cool. For some reason, I'm not playing the computer. It's just like me versus me, which is kind of annoying. Uh, but it did beat Claude on speed. Let's see how Claw performs in terms of the actual game. I would actually say in some ways, you know, I think the game that Chat GPT created was a bit more interesting. Let's just go back to Chat GPT here. Look at this one. This is way more fun. No. So, I mean, in terms of the output side by side, I would say that GPT 5. 4 created a better game. Not that I'm recommending switching or anything like that, but it's it creates something pretty nice. It's done a decent job there for sure. Oh, this is interesting as well. The zoom says, "I'm impressed with GPD 514 in computer use. Absolutely fast and precise. " Nice. Yeah, it's done a pretty good job there. It's created nice for sure. Let's try

something else now. So, what I'm going to do is I'm going to take the information from my website here and we'll, you know, this is an actual use case. So, we're going to say right take the information from my AR profit boarding website and create a new redesign. Let's see how it performs when it comes to creating a new website for my AI profitable community. So, it's beginning to do that right here. Is this AI or the real person? This is a real person. So I've just plugged the copy right in from my website plus this prompt on redesigning the air profitable website and then it's going to start creating a new version here and we'll see if it's better than the original. The original was actually built directly with claude. So be interested to see okay which one outperforms each other. Let's have a look here too for computer use. How are people using that? Is that just through I assume that's through the agent unless you've got your own sort of like browser extension like nano browser for using your computer. I assume people using chat GBT operator to test out. Claude is rude to me. Well, that's no good, is it? All right, so we got agent over here. We could test this out and see how the computer use goes. Um, so I'm going to say, okay, great list on Google Maps of the best Copenhagen bakeries because who isn't looking for bakeries in Copenhagen right now? And we'll test it out and see how that performs in terms of speed on the operator as well. I guess this will be pretty good inside Chat GPT atlas as well. But yeah, it's pretty fast. Look at that. It's a lot faster than it used to be, for sure. So this is the agent being tested inside Chat GPT operator. We've now got the website back as well. So, let's see how that performs. So, it's beginning to create the environment for that. I don't know why Chat Chubity always does this, but when it's designing websites, it always seems to have like this bit over here. Just doesn't look good on web design. Should be like cleaner. Like this should be the main sort of banner and then that should be underneath. Like if you scroll down the page overall, it's not bad, but that doesn't seem like a huge step up from the designs I've created with codeex before. Certainly Sonet beats it in terms of the design and the UI that it would normally create. So I would say Claude is winning in that respect. The other thing I'm not 100% sure of is if we're using operator. Let me just double check here. Yes. So if you're using operator, it doesn't say that it's implemented inside. So there's no confirmation that GPT 5. 4 is inside uh operator. Chatg. That's what I'm wondering. Although it does seem to have become a lot faster in general. But overall, it's one of those where I don't think I would switch anytime soon, but it seems pretty decent, right? It seems like a decent start. I just think that Claude is so far ahead at this point, like it's almost impossible to catch up. So, just to recap on GT 5. 4

literally just came out. It's built for work. So, it's designed to be less errorprone, more efficient, and better at workplace tasks like drafting documents. and we can create files in fewer tries and it outperformed office workers 83% of the time which is pretty crazy. It uses fewer tokens and it's faster so it's faster and cheaper and then also it can control your computer. On top of that it can work inside Excel and Google Sheets as well. And there's two versions that I can see of it which is GPT 5. 4 thinking which is you know the version um that we've been using for creating the websites today but then you've also got GT 5. 4 4 Pro, which is probably super powerful and amazing for research, but it just takes too long to test on um videos like this. And that's

basically it. So, if you want to get all the video notes from today and all my best trainings and everything else, you can get that inside the AI profit boardroom link in the comments description or just go to the AI profitable. com. This is my community focused on helping you save time and scale with AI automation. Inside the community, I have 2,600 members, which means that you can get help, you can get support, you can share your wins, etc. And there's always people online, which is awesome. Inside the post here, you can see that we have a daily accountability group. So, if you struggle with information overload or overwhelm, you can post your goals, stay accountable, and stay laser focused. With this, you can also see what other people working on, which is pretty cool. And then also, each week I condense my 60 hours of research into a quick weekly update that tells you what's good and what you can ignore so that you can save time um and just focus on the good stuff. In addition, we have four weekly video coaching calls, which means you can jump on live coaching calls, get help and support real time. You can meet people in your local area. So, you can just zoom into your local area and find people um locally. We actually have at this point 134 pages of testimonials of people winning from the school groups and the trainings that we do. So, there's a lot of people getting awesome results, which is great to see. And then also inside the classroom here, you'll get our best training. So, for example, you can go from beginner to expert with AI automation and learn how to build your first AI agent in under 5 minutes. Additionally, you'll get all my best trainings, for example, for how to automate X, newsletters, shorts, Instagram AI avatar videos, etc. And then additionally, you can get my daily updates here, which means that every single day where I did new guides, training you on new things, I look at all the updates that came out and I give you the most relevant ones and then I share them with step-by-step guides and video tutorials inside the AI profit board. So, it's always active and is always coming out with new trainings that are updated. Additionally, you can learn how to get more clients with the agency course. You can watch back the coaching course if you don't make them live. You can learn how to rank number one with AI SEO inside Google and AI search engines. And then also you can learn how to grow a YouTube channel based on what's working for me with AI directly inside the classroom as well.