GPT 5.2 vs Gemini 3 Pro vs Opus 4.5 vs Grok 4.1: Best AI?

GPT 5. 2 versus Gemini 3 Pro versus Claude Opus 4. 5 versus Grot 4. 1. Who wins my friends? Today we'll be testing it live. So right now four AI models dropped recently. We've got Claude Opus and we will enable extended thinking. We'll be testing this stuff out. We have Grock which we'll use on expert mode. We'll also have DT 5. 2 on thinking mode testing today. And then Gemini on pro as well. Don't ask me to use deep think because that takes way look too long. Ain't nobody got time for that. And we're going to get straight into this.

So, we got a bunch of prompts that I'm actually going to test out today. And we're going to kick it off with something pretty simple, but I want to see which one can handle it, which one can't. All right. So, we're going to say here, make an animation of a 2D duck riding a bike in HTML. I'm going to click on canvas and submit over here, which is the same inside chat gbt. We'll run it inside gro as well. And then we're also going to run that inside Claude and we'll see which one performs the best. Who have we got on the live stream today? So we got Hanzo. Hanzo says, "Are we playing? " We are playing indeed. Who follows Julian on school? I do. Hands up. And then Technob Ben says, "Hi. " So hello Technob Ben. Going to keep going through this now. So we've got Claude Opus coding out. As you can see, Grock is working its magic. In fact, Grock has already finished, which is pretty cool. So we can just play that in a sec. All right, we have chat GBT 5. 2 and that's still coding now. And then we have Gemini 3 Pro. Now, out of all the tools that I've tested so far, I personally prefer Opus 4. 5 and also Gemini, but let's see what we get here. Gemini is giving us something really cool here. As you can see, this is a dark red in bike. We can change the speed at the bottom. The output looks really cool. And I like the vibrancy of the colors like the clouds, the roads, the parallax, the way we can speed it up and control it looks really cool. If we have a look at Chachi, that's still can. So, we'll give it a bit of time. Now, let's have a look at Grock is I don't even know what this is. This is not on the same level as Gemini. As you can see, I would put Grock in last place for this particular one. And then we've got Claude Opus here as well. Let me know in the comments as well which one you think is pretty cool. This one is interesting. We got Claude Opus duck on a bike. It's got the title at the top. Then we have a duck on top of a bicycle. I think I'm going to say Gemini is winning so far simply because he can control the speed, right? So, I'd say Gemini is winning. Claude is really cool though. It does look good. Let me know in the comments which one you prefer. And then Chachi BT is the slowest model by far. As you can see, it's taking quite a long time. We're now going to preview it. Oh, here we're talking. Wow. So, we can wobble the bike. We can pause it. We've got more control. We can change the speed. Background is pretty cool. This kind of seems a bit buggy, but I like the fact that you can control it and you can mess around with this and that sort of thing. Oh, honestly, I'm going to put that number one. That has surprised me. Chat GBT 5. 2 seems to have created the best output so far. Gemini has created something pretty cool. I do Gemini, but you can see it's like basic like he's just stuck in the middle. He's not really going anywhere. The 2D bike with this is pretty cool, but there's no controls, so that's why I'm grading it as number three. And Grock, I don't even know what this is, mate. I don't even know. This super basic, right? I wouldn't say that's a great output at all. Atif Goldie Sports here. Good to see you, Sporus. Thanks for joining the live stream. And so far, that is test number one. All right, let's kick off with the next example. So now we're

going to say code PS5 controller in HTML. We'll run that. You know, I might sign in to Grock and just see if that works better. Let me just sign into Grock here. See if we get better outputs. All right. So I'm going to switch to Grock and we will test code a PS5 controller in HTML. Ricky says, "Nice to see the real Julian always. " Chat GBT may have won. Was that his arms? I'm not sure as well actually. You never know with this AI. You never know what he's coming up with. And have we got we're thinking on here? This is 4. 1. I think this Okay. Yeah. 4. 1. All right. So, let's test this out. We have the PS5 controller. Actually looks pretty cool. It's got the logo inside there. Buttons work. They're a little bit messy as you can see, but it's not too bad. We can't click R2 and L2. We can't click the directional pad. So, the only things that are clickable are really these buttons here. It is very messy when you look at it cuz like for example here, triangle is over here, but the clickable triangle is down there. Doesn't quite make sense. I'm not that impressed, but we'll see what we get back from Jack GP 5. 2. So, this is 5. 2. You can't really click anything here. It looks a bit messy. So, you can't like click the buttons or anything like that. So, I'd actually say that's worse than using Grock. At least with Grock, you can click the buttons. 4. 1 as you can see. We go over to Claude Opus. Not sure about that, mate. It's clickable, but look at that. What are L2 and R2 doing over there? Not happy with any of these designs, to be honest. And Gemini Pro. What is this by Gemini Pro? Let's have a look. You can't even I would say that's probably the worst one, right? Gemini comes in last right there. And let's just document who came in here. So on the first test it was chat GB 5. 21 Grock last Gemini claw 4. 5 open. All right. So for the duck challenge you see the results. And for the PS5 controller challenge I would say probably Grock did the best. I wasn't happy with any of them to be honest. Claude pretty bad but at least it looked like a PS5 controller. Charge 5. 2. Yeah it's not even got L2 and R2. And then Gem don't know like this is not a PS5 controller guys. All right. So Claude CH GBT 5. 2 too. And then Gemini came in last right

there. Simple camb web app. Let's do it. All right. So, next challenge. Let's test this out. Thank you for the idea, by the way. If people want me to test something else, just let me know. So, build a simple cam board web app. I'm actually going to start a new chat just see if we get better outputs when we do this. So, let's plug that in. Do the same inside. And we'll do the same inside Gemini Pro. We're getting some good suggestions through by the way. So, thanks Vic Gaspar as well. We will build a website. So, we'll come on to that in a second, don't worry. and we'll test out the simple camb board web app at the minute. All right, so we're going to test these out. See which one we get the best back from. Switch canvas on there and we'll see how this goes. All right, so we'll play. Let's see what we got. We can add in a task. If you click add tasso, it doesn't allow us to do anything here. All right, so I'm clicking add task and it's not doing anything, which is not ideal. Let me just change my site settings here. We've already allowed everything, I think. So it should be fine, but it's just not working. It's failed on that completely. All right, let's have a look what we got. So, add a new task on here. Test new models. Add it. Yep, that's okay. Let's move this down to in progress. Yep, that's working fine. I would say that's working perfectly from Claude Opus. So far, Opus has smashed it. GT 5. 2 is still coding out. And then Gemini, let's see what we got. See, this is more what I was expecting. For example, here you've got a sort of side view where you can drag stuff side to side. That's much more like a camb board. If anybody uses Trello, you know what I'm talking about. And so you got todo in progress and done, right? And you can drag these in. And then you can add a new card here. All right. So we'll add that in. And you can see here that Gemini came in number one so far. Qualopus came in number two, but obviously it's not side to side. So that's annoying. And then Chachi 5. 2 is still coding and is pretty much coming in last because he's created something you can't even use, right? So let's see what we get right here. Vasper says, "It's amazing how quick and accurate your mind at typing is. " Thank you. Actually, someone was watching my typing the other day and they were like, "Wow, that's crazy fast. " And to me, it just seems normal, but I think to everyone else who doesn't use a laptop all day, like it's it seems weird. So, Rick has put his ratings here. Gemini, Claude, Chat GPT, Grock. Yeah, I would agree. We've just got to wait. Let's test out Chat GPT. See what it's got. Maybe it'll perform better than Gemini. We'll see. This is looking nice, isn't it? This is like a proper app. It's beautiful. Do you know what? I think Chapter 5. 2 has got a lot better coding or something cuz this looks super nice. It's actually surprising me and it's coded some nice stuff out today. Right. So, this works perfectly. We can edit stuff. We can delete stuff. So, we can click on delete there. We can save that. Move stuff around. We can add a new card in. Yeah, that is top draw. All right. So, I'm actually going to put Chad GPT 5. 2 as number one here. Then I would put Gemini coming in number two. We have Claude coming in number three. And Grock is failing at everything so far. All right. The camb board challenge GPT 5. 2 came number one. Gemini came number two. Claude came number three and came last. All right. So I saw there was this is interesting. All right. So we got some suggestions. Now this is what I love about the live streams. You guys are always getting involved. This is what we like to see my friends. So I know that Vasper asked me to build a website. Neuroplasticity. Thanks so much for reposting the live stream. Appreciate you. Ch seems closer if not better than Gemini. Like the UI is super nice right here.

So, build a modern dark mode personal portfolio website. Okay, I like that. All right, so we'll test this one out. Thank you. Great suggestion there. So, I'm just going to type out this prompt. All right, we'll grab that prompt like and let's see what we get back. And by the way, if anyone has anything you'd like me to test as you go along, feel free to ask. That's what we're here for, right? That's why we do these live streams. So, now I'm going to plug this in. One thing I think that's important here is just starting a new, if you're going to test this stuff, just start a new chat each time. So here we go. Let's go. We have Grock as well and we'll see what we get back. So what we thinking here, peeps, at the minute, chat GP 5. 2 seems one of the best. I would say Gemini is right up there. Opus seems to have got worse at coding since I last used it. Like it used to be amazing for coding. So I'm interested to see what you think is going to come first, etc. Just whilst we're doing this well, like some people will say I need to stick with one AI model to master it. And people say they need to master one model before trying others. But that's like saying you need to master a hammer before you're allowed to use a screwdriver, right? You don't really master tools. You match tools to tasks. And the real problem is not a lack of mastery. It's like not understanding which tools you need. You might think like deep knowledge of one model will save you time. But the model you picked might be terrible at the exact thing you need today. And a carpenter doesn't use a hammer for every job. He grabs the right tool and he gets to work. So the new belief I recommend is like test fast, pick winners, move on. Right? So when you start a project, spend five minutes testing the same prompt across three models. Use the one that works. Done. The best AI users don't know everything about one single model. They know which model wins for which task. And that's what I'm trying to show you today. So, going to go back to chat GBT now. That's coding out. We should have the outputs back from Gemini. Yeah, we do. All right. Let's have a look at this. This is looking super nice. I'm liking that. I'm liking that a lot. Nice. It's missing out some icons here. So, I'd recommend like it has some icons instead of just a dark background. And then it has the contact form as you can see right here. Pretty cool. So that's looking good. Let's have a look at Opus. Opus is pretty good at coding websites. Everyone have seen it in the past. Come back to us as well. J 5. 2 is still coding out. And let's have a look at Claude. So here's something that is a bit annoying about a lot of these different tasks that we use. Right? So if you have a look here, we've got Claude and you see how it creates white text on a white background. Like why does AI still do that? I don't understand it. This could be the best design in the world, but if I can't read it, what is the point? So, Claude Opus is letting us down today. We have got extended thinking mode on, so it should be a lot better than usual. I would expect it to be one of the best honestly, but today it's letting us down dramatically. So, Gemini is coming in first so far. Let's preview the outputs from Chat GP 5. 2. And I know I think Chat GPT 5. 2 has been cooking here. This is looking good. This is looking very nice. Look, the navigation bar is working perfectly. Feels a lot more complex, a lot more interesting. I like the animation that it's got right here with the design. We can go on the contact me form and click directly here. So, all the hyperlinks on the page work and then you just fill in the gaps with like your name and stuff like that, right? And then it's got links to the contact information. This really feels like a developer website. I don't know what you're thinking in the comments, but I would say that actually beats Gemini so far. And this is shocking today. From what I've seen, Chachi 5. 2 has become like a beast at coding. I've never seen this before. So, fair play to it. I was saying the other day that I wanted chat 5. 2 to get better. And I think Big Sam's been listening, hasn't he? He's been watching the live streams. He's been lurking in the comments. I don't know what's going on, but something has happened. And chat 5. 2 is way better than it used to be. And then if we have a look at Gro here, let's preview this out. Oh, this is looking at that. Messy. We don't like that. You can't see the icon. I was liking this. This looks really cool, but then this bit is looking nasty. We don't like to see missing images. Contact me is okay. I would say that's a very average website for where we're at in this world. It could be worse. You could do white text on white backgrounds, mate. That would be even worse, wouldn't it? Who would do that? What sort of AI would do that? Border says Gemini may have an edge due to the color. I was thinking that, but then I was thinking like it actually looks pretty stylish the way it's done it. I reckon it's purposely gone with a black and white theme cuz it's almost like dark mode. It feels and looks like a tast. So yeah, I'm going to go with that. He's definitely watch your live streams. Yeah, I think he's Why is he not in the AR profit boardroom? That's what I want to know. Get yourself in there, Sam. Get yourself posting your wins inside the AR profit boardroom community. From what I've seen so far, I'd say that Chad 5. 2 absolutely crushed it. Gemini did pretty good. Grock was good up to the first bit and then it got pretty bad. And then Claude Lopus delivered something totally unusable, right? I just couldn't use that. And so, if I had to rank these, here's what I'm going to say, peeps. Jack GP 5. 2. Look at this. Look at the rankings. 5. 2 5. 2. Right. Like it's absolutely crushing it. Then you got and then you have Claude at the bottom. Before you go change the color on Claude. I will, but I just I feel like that's cheating. If Gro gets two goes and nobody else does, then we're cheating, aren't we? But I will give it a go. So, I'm going to say let's change the color here. Change the color so you can read the font, etc. And then we'll go with that. In the meantime, let's start thinking about the next prompt. Just want to see what it came up with. Yeah, fair enough. Fair enough, mate. So, it's changing that now. Is the co-pilot GPT smart option a good code assistant choice? I think honestly if you want a good code assistant, you probably want to go with something like I think cursor is really good. A lot of people say good things about it. I also like anti-gravity. It's good if you want to do basic stuff. BS code has always been pretty good. Never really used Copilot honestly. I don't know. I just I didn't like the UI and I felt it was a little bit clunky compared to everything else. So that's improving everything on the page now, but it seems to be taking absolutely ages. So maybe we'll just leave that running in the background and we can come back to it later. All right.

Got to go with the snake game. All right. The classic snake game. Why not? All right. Should we do more of a twist though? Should we make it more interesting with the snake game? Give me a fun dope inducing prompt for a crazy snake game that's harder to code but super interesting. And also give me the prompt for it. You got to jazz it up as supporter says. That's what we're talking about, mate. If anybody's watching this and wants me to test something else out as well, just let me know. All right, the neon serpent gravity shift arena, not snake. All right, so we're going to plug that in. And here we go. Snake jazz up. So, these are beginning to code out. Let's see if we got the new version of the portfolio website, but it doesn't look very good. Let me refresh that. Yeah, we refreshed it and it still looks pretty trash. All right. Anyway, we'll X off that. Let's get back to the main challenge right here. They're all taking their sweet time. Claude, ever change your colors? No, so far it's pretty bad. Mr. No sir, I'm currently working on a super app and using copilot GBT and I think it's so far but would also be willing to follow your advice on such product on which will be the best thing you can do right for stuff like that. I would find out and I would find other people working on similar stuff like this. Join the air profit boredom community right and post this idea just ask for help. So a lot of people what they do is post inside the community here with the projects they're actually working on and then they'll share stuff and share how it's working etc. Right. And so they actually everyone like posts. It's a super active community and people actually comment on there and give feedback, help, advice. Also, you can share your wins, share what you've created. I think that's the best way to do it. So just share that stuff inside the profit boardroom and then you can get help and get support and advice and that sort of thing. Unrelated, but how can I get Gemini or GP to make asphetic sheets? Just ask it. Go inside. For example, I use Claude a lot and I'll just say create a beautiful dashboard and it will generate a really beautiful dashboard for me. The same for example Gemini. You can create like nicely designed spreadsheets, dashboards, SAS tools, etc. So, however you want to present the information. If you just want to focus on Google Sheets, that's okay. And then just use Gemini and ask it directly inside the chat. So, we got Neon Serpent ready to test. Let's have a look what we got here. That's looking all right. As you can see, wa wow. Okay. I don't know if you can hear that, but that was pretty noisy. That was pretty noisy, my friends. I'm actually I'm going to mute that and we'll test it out again. Actually made me shake. Here we

go. Boom. Shackle. It's a pretty cool game. Way more interesting than normal snake. There we go. All right, that's looking awesome so far. Let's have a look what we got. So, Chad always seems to be the slowest at coding, but that's okay. If you give us the best output, it's worth the wait, isn't it? QWE did not hear. That was cool. Yeah, it looked really cool. And Nahome says, "I'm building Microsoft with tenantbased system. Can you guide what I should do

for the back end? Yeah, I would use superbase for the back end. Something like this. And then yeah, Superbase is pretty good. Then you can use like the Stripe MCP as well for getting customers. Those two integrations work pretty well when it comes to launching SAS. And then I think in the modern world like everyone can create SAS tools. So really one of the biggest things is just distribution, like learning how to really market it to all and to get the audience in. So we've got

the neon serpent from Claude. Now let's have a look what we got. Actually going to mute this as well, just in case it gives us some crazy stuff. Woah. That seems to break straight away, doesn't it? What on earth is going on there? That seems impossible to play. All right, this doesn't seem to work. What's going on here? That's what it's giving us back. That is nonsensical. So, at least Claude gave us something back, although we can't play it. Gemini gave us an absolute beast, as you can see. That looks really cool. That's all we wanted, guys. That's all we wanted. The best AI in the world can do it. Interesting. So, let's have a look at chat GP 5. 2. We'll preview this. It's looking good. It's looking promising so far. Wow. Okay. What on earth? What's going on? We'll go on chill mode. Chill mode is not that chill. This seems unplayable as well. Yeah, I would say that's unplayable. It doesn't make any sense. Claude is underperformed today. It's not reached its full potential. I'm going to put Chachi 5. 2 number two here, but I don't really like the game it's created. It's not that great. Claude was unplayable. Grock was unplayable. Didn't even create anything. Gemini created an absolute beast. Absolute beast, mate. Fair play to it. Look how much fun this is to play. And I've got to say like Claude so far today has not really been in the race. It's really struggling. So the jazzed up snake, we had Gemini, ChachiT, Claude, and then Grock. So those are the ratings so far. Chat GPT the UI looks super nice, but it was an unplayable game. Like literally the front end looked nice, but the actual playability of the game just doesn't make any sense. So if you want me to test something else out as we're going along, feel free to ask. Otherwise, what I'm going to do is just keep going through with some of the tests that I've built and some of the stuff that I wanted to test out here today. And these ones are a bit more crazy. So, we're going to plug in a new chat. And for this one, we're going to say create a retro wave endless driving experience with a grid based road, palm trees, blah blah. Claude Opus has let me down today. One of my favorite models and it's underperforming here. So, we'll wait for this to roll out here. Now

some people say I can't test multiple models cuz I don't have time. People say they don't have time to test, but testing only takes 5 minutes. And fixing broken code or the wrong model takes 3 hours. So I would say no one's too busy to test. You're too busy because you don't test. The real problem is like not a lack of time. It's frontloading fear. Testing is work prevention. Because 5 minutes of testing saves you from hours of fixing, rewriting, explaining to clients why the tool doesn't work. Test fast, move fast. I recommend that your new rule should be that before you commit to building with one model, spend five minutes testing your exact prompt across three different models. Pick the best one. Build once and do it right like you can see today. And it doesn't take long to test this stuff, but it will save you a lot of time in the long run. Bear in mind like things are always changing, right? So this is the 3D

driver from GR as you can see. It doesn't look that impressive. Doesn't look that interesting. It seems almost unplayable. Claude is still coding this out. We have chat GP 5. 2 still building out. And then Gemini is always like the fastest here. But again, like it doesn't really make sense. Like what even is this? I wonder what JP 5. 2 will do. Claw I'm not holding out for. I don't think it's going to be great. But actually, having said that, it looks better than everything else I've seen so far. So this is the tool. Yeah, I would say that's the best so far to be honest with you. Gemini, I don't even know what that is. Grock, I don't even get what this is. That is not a car, mate. And then Chy 5. 2. We'll just wait for that to finish up. You know what's really interesting is like when Claude Opus first came out, it was absolutely insanely good and now it's very average at best. So, Rkitio says Claude probably allocating resources to enterprise clients now that the workday is here, but that's pure speculation. Interesting. Vince says, "Are any of these multiple AI subscription sites legit? " Yeah, they are usually legit. I don't know any that are like a scam. Yeah, I mean they're usually they're pretty good. you won't get like the full power of these other AIs. You can use them and they're pretty good. So yeah, if you want to open router is the best for that. If you just want to go straight to the source, just go open routter. Sporter says, "I've just changed to GPT 5. 2 in Windsor. Opus was stuck all night accessing my terminal. I think something is wrong today. " Ah, that might be right. Claude looks okay, but it doesn't seem like the car is move. Yeah, I would agree. So, we've got some a pretty bad bunch right here. Claude 5. 2 still coding out. Yeah, I do wonder whether there's something wrong with Claude Opus today, but let's see. So now we can preview it. Stop. That was loud. That was very loud. I don't know if anyone heard that. I actually think with chat PT 5. 2 is probably one there. It's a weird game. Like I wouldn't play it fun, but that actually works. It actually got car. Do you know what? I'm actually There's tie, isn't it? It's tough. Which one do you think? I'm going to let you the people decide here. Do you think the chat GPT 5. 2 one, which is this one, or you think that Claude created the best one? Which one do you think is the best? Let me know. Oh, yeah. Perplexity is pretty good. If you got a Plexi, I think you're good to go. Like you don't need to switch around there. Plexi is one of my favorite tools to be honest. It's pretty reliable. Sporter says low. No one's voted. So I'm going to say the chat GPT. I do think it's probably created the best. We got a tie though. Sporter says chat GPT 5. 2. Nom says Claude. I'm saying Chat GT 5. 2. I'll give you one last shot. All right. Yeah, I'm going to say Chat GP 5. 2 then. So retro driving game chat 5. 2. Claude Gemini and then Grock. Gro has come consistently last from what I've seen. You can Gro is last. The last not the best. Wick says just got here. Claude looks better. Does the GPT function better or something? Yeah, it functions better. It's a little bit less buggy. A little bit less. Yeah. Yeah, I would say that Claude created a better output right there. Sorry, GP. What do you think Grock is good for? I'd be interested to know what people watching here think Grock is good for. I'd probably say like creative stuff, for example, like creating tweets and that sort of thing. Everything else I've seen it do today has been pretty average at best, if not really terrible. I'm going to try something else now. We're going

to say create an interactive 3D aquarium using HTML. Let's mix it up a bit. Give good old Gemini cheeky. And we also have Gro running in here too. Mr. Note says as we continue to get enlightened on AI coded assistance use, Goldie, I've got a question related to what I asked earlier. So for a super app to have a separate admin panel front end from the super app's front end is a website or a separate app the best choice to build for the admin panel or the work I'm using AI code assistance development tools is fiber by you of course yeah so the one that I've used before previously was data button was pretty good for like backend front end you can use like lovable and bolt as well for mini apps pretty good and then you just combine it with something like super base and just build out all the different elements of it I think you can't go wrong with those sort of things I've never really built like a fullblown super app offline or locally for example with VS code. So the one that I've built probably the biggest app with was a job board and I built that job ad with sorry job board with data button and it was pretty good like I was actually quite impressed by the output so that was the best and then I've built stuff like for example basic apps with love and bolt but just the front end not the back end as well and then cursor as well is pretty good for creating the back end. So I would say go if you want to build something based on my experience then go with either data button or cursor. You can also use a replet as well as the other one that people like a lot. So we're going to preview this output from chat 5. 2. It doesn't look like it's working. Click to feed the fish. I'm clicking and I'm not feeding any fish. Let's try Gemini. Enter the aquarium ammo. That's looking beautiful, isn't it? Even though there's an error, which I'm going to fix now. That's actually looking pretty cool. Grock says click to feed the fish, but we're not feeding anything. Gemini has actually created something pretty cool. I really like what it's done there. Five is still coding out. And Claude Opus is loading this up. Oh, this is good. This looks good. That looks like a proper aquarium. So, we're going to feed the fish here. We can drag it to rotate it. So, we can rotate it up and down. We can click to feed. I'm absolutely loving it. Absolutely loving it, mate. All right, we go back to CH G 5. 2. Let's see what we get back. Rock obviously comes in last. Gemini is fixing the error, so we'll give it a whirl. Claude the Redeemer. Yeah, that's exactly it. And then we'll enter Aqua Sim 3D. And we're in night mode here. I don't really like night mode. Oh, what's it doing? I don't like that. It's flashing away. Just gonna click off there. Sorry about that, peeps. Oh, I think that's broken. Gemini is broken. I'm not even going to click on that. It was flashing like crazy. Like a maniac. Brock and Gemini both failed on that task. Claude the Redeemer created something really nice. This is pretty awesome. You can zoom in as well. Look at those fish. So realistic. We can feed the fish as well. We got the temperature, the pH level. It's just a beautiful what a time to be alive. And we're running out of actual session space on Claude, which is interesting. We'll just wait for chat 5. 2 to come back. And then we got one more test that we're going to do here in a minute. So we got chat GP 5. 2. Let's see what we got here. We have a bug and then click on fix it. That is not working. I clicked on fix it, but it didn't do anything. Okay, so that totally failed. I'd actually say that's worse than Yeah, that was worse than actually using everything else to be honest with you. At least Gemini gave us something that we can actually. All right. Here's what I'm going

right. All right. Here's what I'm going to say then. Chat 5. 2 came in last. Grock came in third with an unusable thing, but at least it says click to feed the fish. Then we have Google Gemini, which has generated something interesting, but they're just like blocks, right? There's no detail to those fish. I want to see the detail on the fish. And then we have which created easily the best thing so far as you can see. So, it's really cool. Looks great. Works perfectly. Boob shackle. All right. So, Claude, then Gemini, then Grog 5. 2 too. T says, "How many domains do you own? " Quite a lot. Over 20. More than I can count and keep track of to be honest with you. But I'm probably getting rid of quite a lot just because like you don't need that many domains, right? Unless you're doing something dodgy on the internet. All right, so we're going to now type this out and this will be the last test, my friends. And then we'll count them up and rank with AI keyword in. Oh, only a few, but we've created a lot of pages. Gro, great for annoying lefties. And then Gary says, "Best video AI model so far right now according to you. " Yeah, Vera 3. Very 3 is the best for videos right now. So, we're going to take that and we'll plug this in. Sporus, you're a legend for giving us the prompts today. Appreciate that, my friend. And we'll run this through. Make sure we got thinking on and plug that in. Oh, look at that. We've reached the limit. Claude has failed. I haven't even used Claude that much. I'm going to wait until December the 22nd at 7 p. m. All right. So, actually Gemini actually created the image, Nana Banana image, which I didn't want to do. Did someone say creator il Hashimo claude is bowed out gracefully got confused we'll rebrief it and then that's pretty crazy from croc I like that but again it got I didn't I should have said in HTML so that's built it out that was nice and quick let's preview it doesn't make any sense that is not what we were looking for so has failed there claude failed chat is coding out as normal wow Gemini created something awesome that is pretty cool all right so hologram when you say Gemini didn't really create anything of substance and then we've got chatb 5. 2 first time here love what you are doing as a fellow AI enthusiast what do you recommend is the best value out of all these models yeah so Gemini has worked really well today I think it's done a great job and chat 5. 2 material has been best. So yeah, I think between those two models, you can't go wrong if you're coding stuff out. Now, having said that, if you just want something that gives you a bit of everything, videos, images, everything else, then I think honestly Gemini, Google Gemini is the best. Reason I say Gemini is because it's got Google's whole ecosystem. You can create images, videos, you can create canvas, you can create flashcards, everything else, and also links to notebook, which is another one of my favorite tools, and I have loads of videos on that if you haven't checked it out. So yeah, I would say go with Gemini, best value. So, let's talk about the old way versus the new way when it comes to choosing AR models as well. I know a lot of people asking like what's the best one. So, the old way that people used to pick was like they would pick based on hype and influence recommendations. They would use one model for everything because it's easier. They would assume the newest model was always the best and they would never test alternatives because switching feels complicated, right? The testing way as you've seen today is that you run the same prompts across multiple models and you compare the results. You match the right model to each specific task. You test new releases yourself instead of trusting all these big claims. And then you build a tested system so you always know which tool to reach for. And this way you get faster results by using the model that actually handles your task best. You deliver cleaner, more professional work. You stay ahead because you know which model just got better and which one has fallen behind. Based on today, I would recommend we'll count them up in a sec. But yeah, let's just have a look at Chachi 5. 2. Doesn't work. It's got a bug. Okay, so they all failed, didn't they? Apart from Gemini. Now let's count them up. So, we're going to take these and we'll plug them into CH GBT and we'll say which AI model won the challenges today. You can see them ranked. So, we'll test it out and see what we get back. Trusted CHP 5. 2 to come back with the best options at placements and low score wins. And the overall leaderboard, so if you come first, you get four points. Third, three points. Second, two points. One point. As you can see, Chad GP 5. 2 got the most points. Gemini came in second. Claude Opus came in third and Grock was nowhere near. Look at that. 25 versus 14. Challenge fights. You can see a breakdown right here. Then the quick read is that Chachi 5. 2 got the most wins plus the strongest consistency. Gemini got the most second places and two big wins. Claude won with the aquarium win but had lots of solid mid placements. And then Grock had more force overall. All right, so you can see how this works and you can see which one performed the best across each task. And as you can see, chy 5. 2 won it today. So yeah, embarrassed to say never heard of Google notebook until you mentioned it and I Googled it. If you Google it, you probably just find me and my videos. So yeah, definitely check it out. Notebook is awesome. So you can see the placements here. Chip 5. 2 came in first, Gemini second, Claude for Opus came in third, and Grock came in fourth. All right, pretty interesting to see the results there. Now, if you want to get

access to all the video notes from today, the breakdown, etc., Feel free to get the AI success lab link in the comments description. It's completely free inside school and you can find all the details from today's tests. It's all the prompts etc. All the things that we've run, all the beliefs to break, how to implement this stuff yourself. And if you haven't already, check out the AI profit boarding. This is an awesome community where you can learn, you can grow with us, you can get help and support whenever you need it. If you're struggling with this stuff, you can get lots of cool stuff inside here. If you feel like you're overwhelmed or like you're getting overloaded and that sort of thing, what you want to do is we actually have a focus protocol inside the air for boarding. And the reason for that is because everyone in AI feels out of date and overwhelmed when overloaded. So you can actually follow that ultimate focus manual that I've written myself and you can use that to never feel overwhelmed again. And literally, if you implement this focus manual right here, which we've got a link to inside the air profit boardroom, you will never feel overwhelmed again and you'll never feel overloaded. So, it's all inside the air profit boardroom. Inside there, too, you get all of my best trainings, courses, etc. You can learn how to build your first AI agents in under 5 minutes. If you're a big fan of all the latest updates, you can get all my updates on the left hand side and we actually date them so that you never fall behind. You can see what's new. And then also, you can see here you get all of my best playbooks, courses, templates, workflows, automations, etc. inside the aircraft for boarding and additionally you can jump on coaching calls each week ask questions show up get support etc whenever you need it. The other cool thing that we do here is we have weekly updates inside the prof. So every week what we actually do is tell you what's worth your time, what to skip and we basically condense 80 hours of my hard work into a 5m minute quick update so that you can just skip through everything and get exactly what you need and learn exactly what's good and what to ignore so that you never fall behind. You never feel overwhelmed, right? And then also you can post your wins and share stuff. So you can see inside the profit model here with loads of people posting their wins, sharing cool stuff, sharing what's working for them, etc. And this is all inside the profit boardroom link in the comments and the description. So appreciate you watching. Hope to see you inside there. You can also DM me and ask for help, get support, etc. whenever you need it. And all the country calls get recorded as well. So if you ever need to watch him back or that sort of thing, you get it. Supporter says, "Got to follow Julian. Join the school communities 100%. " And yeah, good to see you, Mark. Good to see you. I'm going to actually going to go now. But appreciate everyone watching and I'll see you on the next one. Cheers.