New FREE GPT 5.2 Update is HERE!

GPT just dropped a massive update, GPT 5. 2, and today I'm testing it live with you right now. This is not a hype video. This is just me telling you whether it's good or not. We're going to be testing out side by side today. This is not me telling you that it's amazing before I've even used it. So, here's what we're doing today. We're going to be testing GPT 5. 2 to against real business stats, comparing it side by side with the old version, running the exact automations solarreneurs need every single day. Finding out what works, what doesn't, and what's overhyped. Building actual business systems live to see if this thing holds up. Showing you the truth, not the marketing claims, and deciding together if this update actually matters for your business or not. So, you can see here, here's the details. Literally just got announced a few hours ago, and they said GP the most capable model series yet for professional knowledge work. already the average chat GP blah blah. Okay, there's months in height and then it says GT 5. 2 sets a new state-of-the-art across many benchmarks including and then it's listed a bunch benchmarks here so you can see the comparisons of GT 5. 2 versus 5. 1 thinking as you can see and if you want to get access to it I actually didn't have access until I upgraded to pro. So just something to bear in mind there is that if we have a look here I do have GP 5. 2 too, but when I was just on the normal plan for I think was it $10 a month, it wasn't there. I had to upgrade to the more expensive plan. I'm guessing they're rolling this out to more people as it comes out. So, just something to bear in mind there as we go through. And you can see here we can start using this for all sorts of different tasks. So, for example, I'm just going to start off with something like let's switch to thinking and we're going to say, okay, code a PS5 controller in HTML. It's one of my favorite tasks simply because like quite often the GP the new models will fail with this even other ones like for example Gemini. We'll test this out with Gemini 3 just to compare it side by side and we shall also test this against Claude as well using Claude Opus which is the new release from Claude Opus. We can just see how chat how GT5 compares against all of them. All right. So when we say code PS5 controller in HTML here as well and we'll see Nazim says there is no end. It never gets boring. The end never comes because there's a new update every single day. So we've got Gemini coding out here and we have the HTML for the PS5 controller inside GPT as well running here. We running a few more tests. Whilst we're waiting for both of those to update, let's see how they compare. So you can see there's quite a big difference in GDP value which is like knowledge worth tasks. You can see for example GPT 5. 2 thinking is 70. 9 versus 38. 8% which is a huge jump but let's see. I mean I don't really pay attention to these benchmarks quite often but simply because in reality things are quite different. Leonardo says hello I do need your help. Please send a message when you're available. Drop me a message inside the AI profit boardroom if you want to get in touch. That's the place to connect with me. So connect with me inside the AI profit boardroom link in the comments description and then you can DM me directly inside there. The one of the worst things about messaging on Facebook in particular is that like it just goes through to the DM request. So it's not the best place to talk. So we're going to have a look in GPT 5. 2. We've got the code as you can see over here. One thing that is annoying is like we have to select canvas if we actually want to run this. So I'm going to say run this code and then we'll start running that. In the meantime, what have we got back from Gemini? So, this is a response from Gemini. So, it's clickable. See, all the buttons on the L2 and R2 don't work. These buttons do. The analog sticks work, etc. Not bad. If we have a look at Claude Opus 4. 5, it's gone super weird. It's done the job to be fair. And these are all clickable. Say it's a tie between these two. Yep. So, just bear that in mind. Maybe it's just rolling out slower to the people who are on the basic paid plans. And imagine the ones on the free. They're probably not going to get it for ages. So, let's click on preview here and see what we got. None of this is clickable. I J LK. Why is the analog stick in the middle? This is a joke. I would say it's failed the test here. If you look at Gemini, at least Gemini is align the analog sticks. All the buttons are clickable. That's a way more interesting creation versus this one which is a total mess. None of this clickable. The analog stick doesn't work. And then if we have a look here, we've got the PS5 controller from Claude Opus. Everything's clickable there. I would say that is much better than Chat G5, even if these weird direction buttons seem to move, which is strange. But yeah, I would say like it's And also it recognizes that we're clicking certain buttons at the bottom. So yeah, I would say like it's failed on the first test. All right, good stuff. So

we're going to switch to auto now. Auto decides which model it's going to use. So whether it's going to use pro, thinking instant. I've never seen it actually auto select pro and then we're going to go into another prompt here and we're going to take this prom that I use for creating SEO content. Right? So I'm just going to take that, plug this in, close that. Then we'll go to GT5. 2. Plug this in. And I'm going to say, okay, keyword equals SEO training Japan. Create an article for me, mate. Right. And we'll select canvas. Let's see what we get. Do the same inside Gemini. So, let's have a look here. Tools, canvas. Let's go with that. And then we'll do the same inside Claude. I'm going to actually I've seen some people rave about Opus writing. So, I'm going to stick with Opus. We'll see which one creates the best. So Eden says, "Oh, so you need pro to access it, but yeah, UI isn't GPT strength, but I love it for building infer and back end. " Yeah, I had to upgrade to get access to this. So that's the price I paid just to share this stuff with you. But yeah, from what I've seen, you it's not rolled out to plus yet, but I'm sure it'll probably come later today or something like that. And so far, from what I've seen, like I probably wouldn't upgrade anyway. Probably not worth it. All right, says all done. Here's what frustrates me about GBT5. Look at the way that it's responded here. And I've seen this ever since GPT 5. 1 came out. So you see how it's got no bullet points. It's weird the way that it writes. It doesn't feel very sentient. And then it's if you want next, we can localize it. It just it doesn't feel good when I read this. Whereas if we go up to Claude, Claude's content is usually way better. And also the way that it just writes, it feels nicer. I don't know something about the way it formats text. It's much nicer inside Claude. Bear in mind that open air did recently trigger a code red because they're worried about their competition, Google taking over, etc. And I would agree. I even see that from the YouTube video numbers. If I create a video, look at this, right? video about Google Gemini, it seems to always perform the best, right? I think there's just a lot more people interested in Gemini than there are inside a GPT or anything like that. So, let's have a look at this one. All right. So if we have a look at the title here from GPT 5. 2 right the title is SEO training Japan. Now that's not really to me that's not an article headline right it's not that interesting I don't think people are going to click through etc. Right now, if you have a look at the title from Claude, Claude is SEO training Japan, the complete guide to ranking in the Japanese market. That's way more relevant, right? That is a much better title for the article. That's something that people are actually going to click through to. And also, it's going to be better for SEO when it actually ranks directly on Google. If you have a look on Gemini, I still don't like the way Gemini, look how generic that is. How to dominate the market without wasting years. Just feels like super generic. It's not even tailored to Japan. and the rest of the title just doesn't feel as good. Claude for me is still winning when it comes to writing. It's a model that I'm switching to every single day. And then if we have a look at GT5. 2 as well says, "Are you trying to rank in Japan but feel like nothing sticks? " Now, I ask a question. That is a question, my friend. Right? And I'm no like English grammatical genius, right? But what I will tell you is that is a question. Are you trying to rank in Japan but feel like nothing sticks? And therefore a question should have a question mark. I would hope so. Anyway, and you can see here that the latest cuttingedge model, the most powerful model of all time from OpenAI, the most hyped model, the model that they put out because they're on Code Red, the model that was supposedly going to save them from Claude and Gemini and all these competitors they have. Can't even use a question mark. It can't even figure out to insert question marks inside the text. How frustrating is that? How annoying would that be to edit? Like you can't just copy and paste that into your website. That's no bueno at all. If you have a look over here, we've got the content from Claude and you see how it's nicely formatted, right? So, for example, it's bolded the keyword. It's nice and easy to read and it just feels a lot more sentient and human, right? And then the same for example on Gemini. You see how it's written and it's kind of like direct response format. It's really easy to skim through. It's like looking for world-class SEO training bound specific strategies and it actually remembers to insert a question mark as well. It just feels better. This is why I'm switching over to Claude for pretty much everything or to Gemini for coding cuz GPT 5. 2 and GPT 5. 1 were some of the worst models that I've tested so far and see like it's pretty bad. Let's duplicate this right now. Right. So, we got one tab with 5. 2. I'm going to run exactly the same prompt and it with the old model GPT40. So this is the legacy model G. I'm going to take the exact same prompt that we used for GPT 5. 2 and I can almost guarantee that the content within GT40 is going to be better. Right? Look

at how that's written. How much nicer that looks. Look how it inserts question marks into the content. Like was a better model than GT 5. 2. Definitely better than GPT. and it's just so much better. Our pair says, "I tried it on Ellen Marina, but I'm yet to have it on the app. " Yeah. So, I'm wondering if it's on Open Router as well for you. Let's have a look on Open Router models 5. 2. Yeah. If you don't have access to it inside chat GBT and you don't want to upgrade to the pro plan, then go to open router. ai/models. We have a lot of training on how to use open routing. And then if you go inside here, you'll see OpenAI GPD 5. 2 to chat, OpenAI 5. 2 Pro. So, they've already released the API for Pro and then GT 5. 2. I think that must be like the normal version, right? But my point here is it's no winner. It's no bueno. Like, we've compared the models, not even against Claude, not even against Google Gemini. GPT 5. 2 is better. Sorry, GPT4. I'm just putting it out there. And if you disagree with me, let me know in the comments because I'd love to know why. Do you have GPT 5. 2 on the website? Yes, you can see it right there. So there's GT 5. 2 and we can switch between these and then we can switch back to these legacy models. One of the most annoying things about this and I haven't figured out how to change this permanently. Yeah, but if I switch to say a legacy model 40 and then I go to a new chat, it's going to switch back to 5. 2. It's like I can't even use the legacy models on default. So yeah, for me I'm switching everything over to Clawude. All my custom GBTs, I'm switching over into claw projects. I just think it's a superior model and it's much better. Let's see what people are saying on X about this as well. I'm interested to know now to see okay how are people doing and what people think about it etc. I love seeing some of this. So Mark says GPT 5. 2 is AGI. How many Rs in garlic? There are zero Rs in garlic. Let's try this out. I want to see if this works or not. I think it's cuz he put R capital Ry S and that's why. Let's test it out. There is one R in garlic. Okay, so who knows? Maybe he tested on thinking on instant. He's figured it out. He's figured out. Okay, it doesn't struggle with that at all. Maybe he's got a different version here. What else we got here? OpenAI thinking evaluations. It's inside GenSpark as well and you can get it on the chat agent. So, actually, if you go over to this a way to get chat GPT for free as well, I guess. Just going to log in. All right. So, if we go over to let's have a look here. new uh you have to be a pro plan to user to use AI chat. So it's not really free is it? Let's have a look here. That's unlimited. Oh, so you can use it for free. All right. So if you go inside if you want to use this for free right now, right? Go to GenSpark. I'm on the free plan. All right. So you can see here I'm on the free plan inside GenSpark. And if you want free access to GP 5. 2, you can get free access inside GenSpark. Right? So, if you want to get free access to this, you don't want to pay for chat GPT5. 2, you want to access it instantly because it's not inside your chat, then what you want to do is go to genspark. ai, then you're going to go to new, then you go to AI chat, and you can get access to this for free. All right, that's how you get access. Agent says, "Interesting take. I absolutely love GPT 5. 1, particularly for code. " Yeah, I think maybe for coding it's better, but like you've seen the coding challenge today, right? you've seen me compare coding a PS5 controller in GPT5. If you give me a test, Edin, in terms of something you want me to test as a prompt that's going to produce better outputs in GP 5. 2 versus Claude and Gemini, just put it in the chat because from what I can see, even when we've done the coding test with the PS5 controller, it's not as good. It's clearly not as good versus Claude and Gemini. Claw code was a mess, but I recently got Gemini and the anti-gravity app is very nice. Yep. 100% I would agree. Pair says I use 40 often. Yeah, I think 40 is probably the superior model inside Chat GBT that they were the golden days of chat GBT looking back when I thought you go back a year, chat GBT was way more in the race than it is right now. GPT 5. 2 is the best model. Yeah, for real world professional use. I'm not sure about that, mate. Specifically, GPT 5. 2 thinking beats or ties top industry professionals. We've tested out for coding today, right? And you've seen that honestly Claude and Gemini still be beat it. Maybe what we can do is I'm going to give it another world, right? We'll give it a cheeky world. We're going to go to here. I'll go into canvas and we'll give it one more shot and I'm going to say create a beautiful modern landing page with direct response copy for my SEO agency Goldie agency as Julian Goldie. funnel them to you booking a free SEO strategy

session and then I'll include the link. We'll give it one more shot. shot because I can see a few comments people asking for it. All right, so this is also frustrating, but we'll come back to that in a minute. I've seen what it's done already. It's no bueno. We'll do the same inside Opus 4. 5 and Gemini as well. Kind of annoying that canvas just doesn't pop up by default, honestly. That's what I like about Claude as well is that I don't have to tell it to use canvas. it just goes off and uses it. So, you can see here Claude is coding the page out. I would expect that to look really good cuz Opus is just goated for creating uh website designs. Like, it's super nice. If we have a look u check GPT 5. 2, you can see here that it's just written the copy. It hasn't actually designed a landing page, which is annoying. And bear in mind with the same prompt, we've got a landing page being coded out with Opus. So, it's not like the prompt is wrong. It's just that chat GPC 5. 2 is not very good at understanding. If we have a look at Gemini as well, it understands that we need to code out a landing page in HTML and then preview inside the canvas. And so again, as a coding test, as a knowledge test, like it's just super annoying. And I could tell it to code. But why would I have to tell it to code? That's the point, right? If I have to tell GBT5 every time I create code to code something instead of it being intelligent enough to interpret that I should be coding it then this is what I'm saying right so chat GBT I have to tell it to code it's not intelligent enough to understand my user intent whereas for example claude understands that I'm trying to code and it codes out automatically without me having to tell it and the same on Google Gemini codes out the page without me having to tell it because It's intelligent enough to understand search intent. So, this isn't about telling it to code or not. This is about, okay, can GPT understand the search intent of what we're trying to do here? And it clearly doesn't, right? I'm going to say code it out. Now, Angger says, which best model for writing research articles? I would go with Claude. Yeah, I think Claude GP 5. 2 from what I've seen today in terms of writing anything, whether it's research papers, whether it's writing content is not great at all. Um, this is coding out now and we'll see what we get back on the preview. But I would say like this landing page is pretty modern, interesting, etc. It's quite a nice landing page. It's written all the content and it's quite long. Even like the graphics and stuff are nice. If we go over to Gemini as well, it's not bad. It does have a little error at the bottom, but it's not bad at all. I still think that the page from Claude Opus is the nicest. And then if we go over to GPT 5. 2, let's see what we got. It's still coding out. We'll come out to that in a second and we'll see what we get back. Always the thing that I would say with all of these models that come out is don't listen to the benchmarks because the benchmarks are usually wrong and usually when you actually come to trying them in practice like you can see with Claude with a chat GPT today, it just it doesn't reflect reality. I think the benchmarks are so far off what you would do dayto-day that it doesn't make sense to even pay attention to them. And I think you've just got to try this stuff out yourself and compare it to what you've already got. So compare it to claw. Compare it to Gemini. See how it performs. Right? So let's have a look at this page from chat cheap. Now it's just like that is not a nicely designed landing page. Right? If you have a look at that, look how messy this feels. It doesn't feel or look like a website, right? Whereas if we compare that to Claude Opus, this actually feels and looks like a website, right? It's quite neat. It's tidy. It's designed nicely. And the same with Google Gemini. Gemini's page actually feels like a website. So that's what we get back from Gemini as a coding challenge. Right? So number one, GP5 didn't even understand that we're coding and we have to tell it manually which waste of time. And then number two, the design from Google Gemini is much nicer. Right? This links directly to the page to booking a call. It just looks neat. It's tidy. It's interesting. It looks cool. Right? that would pass as a landing page. Whereas, if we have a look at GP 5. 2, explain to me what you think it's doing here. Cuz look how complicated it's made it. Look how messy it is, the layout and everything else. I just don't think it's anywhere near the same level. Just to recap, peer peeps. All right. If you want to get free access to chat GPT 5. 2, you don't want to upgrade, then what you do is you go to genpark. ai. You're going to go to new chat and you can type in here exactly what you want. Right? That's how it works. If you want to get the APIs, then you would go over to open router, right? So you can get the APIs inside open router. You go to models and you can grab all of these. Right? If you want to know what apps you can use the API in, here's an example. So kilo code for example, client root

code, all the usual tools that you can use, right? You can get an API for the pro version as well. When we've tested out GPT5. 2, it's performed worse than writing content with GPT40, it struggles to understand user intent as well as Claude or Gemini. He's failed all the coding tests versus Claude or Gemini. So, it came last. It couldn't write in a question mark on the content that we produced and it doesn't write headlines for articles that we produce. Right? I've just paid $200 to upgrade on chat cheaps so I can get access to it for you. And you can see the problem here. I really do hope that chat GP and OpenAI turn it around. But from what I'm seeing is it's not even as good as GP. I feel like we've gone a year backwards with the models. So let's have a look what else is said now. So workforce planner create a workforce planning model headcount hiring and attrition include engineering etc. And this is GPT 5. 1 thinking versus GPT 5. 2. So this is formatting in spreadsheets and slides. Apparently it's better in coding although it seems to have got. And it says GPT 5. 2 represents the biggest leap for GPT models in a genic coding since GT5. And I just I'm not seeing that with the outputs I'm getting. Maybe it's because everyone's using the API and usually when these APIs come out they're not that great. But from what I'm seeing I'm struggling to get good responses with it. It's just not that good. And I don't mean to slay feel a bit harsh on it today, but it's just what are we doing here, guys? Why are we creating models that are worse than GT4? And then we can see the benchmarks here, right? It's quite a long article. So, that's basically it for GPT 5. 2. We've tested it. We've compared it. We've seen which one performs the best. We've given it coding tests, writing tests, and everything else. If you want to get the video notes from today along with a community of 38,300 people who are all interested in AI, you can see it's a super active community as you can see right here. Then you can get that inside the AI success lab link in the comments and description. It's a free community with 38,000 people you can connect with. It comes with all of these different lessons, NAN automations and everything else. Link in the comments and description to that. We also update this daily. So if you go to the classroom and then go to more AI automations, you can see we've got the video notes from GT5. 2 today along with everything else. And then if you want to get help, support, coaching, if you want to get all my best systems and connect with me personally, then join the AI profit boardroom. Inside the AI profit boardroom, we have an awesome active community of serious AI builders and entrepreneurs who are growing their business with AI. So you can see for example here loads of new cool posts, lots of people sharing cool stuff, etc. And all of it's inside the community. So feel free to get that link in the comments description. And inside here as well, you get all of my best automations. So, a lot of people ask me, for example, like how to create AI avatar videos, and you can learn that inside the AI profit boardroom. If you go to the AI avatar section here, we've got loads of training on AI avatar videos and how I create them. You've got, for example, my training on shorts, newsletter automations, Twitter automations, and also we do four coaching calls per week, right? So, you can jump on these coaching calls, connect with the community. They get recorded as well if you need to watch them back. And one of the other cool things about this community is that you get unlimited technical support. So you can ask questions inside the community, get help, get support, etc. whenever you need them. All right. Damicious said, "Did you see they added Photoshop into chat? I've actually got an avatar video coming on that later today. " So yeah, we've got that inside the community. We also have a daily accountability post. So you can see here for example every day Darren and another legendary member Johan they both create these accountability posts and then we share them inside the community here. So link in the comments description to join the AI profit boardroom community. It also comes with like tons of NA10 trainings and everything else. An agency course on how to scale with AI automation and land AI automation clients. You've got a full YouTube masterass right here. the YouTube AI road map that shows you how I grew my channel and my best AI SEO automations, too. Plus, we update this with new SPs as you can see inside the classroom here. So, all the latest new SPS as you can see right there. All right, link in the comments description to that and thanks very much for watching. I will see you on the next one, peeps. Cheers. Bye-bye.