Grock 4.5 Review: Testing Against Claude Opus 4.8

All right, we have a brand new update from Grock 4.5 today. I'm going to be building out into the agent OS. So, you can see the announcement from Elon Musk over here. Grock 4.5 is out now. It's not using the new internally developed in software, but it is out right now. Built on the V9 foundation, scoring pretty well, just behind Fable 5, Gupty 5.5, and Opus 4.8. 8, but matching GPT 5.5 on coding benchmarks at lower

costs. Only way to know if it's good or not, my friends, is to make sure that we test it on the benchmarks. So, we're going to get this party started and we're going to start looking at Grock 4.5, building it out, and also testing it on Goldiebench. See how it performs versus everything else. So, let's see how it does. So, if you're not familiar with Goldiebench, this is the system that we have. We we usually test

about 47 different tasks with models that we're looking at. As you can see right here, and we need to add Grock 4.5, which I'll do in a second, and we'll see what we get back. So, let's get this party started. Have a look. Now, if you got any questions whilst you're watching, feel free to ask. If not, I'm going to get straight into coding. Heat. If Twitter if Twitter actually loads. Let's check that out first. I

think the internet's really slow today. Wow, look at that. might have to switch Wi-Fi a It could sound bad. Heat. Heat. N. What is going on? Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Let's check the SEO projects. See how we doing. If you got any questions, by the way, whilst you're watching, feel free to ask. Happy to help Heat. Heat. DG Jerome, welcome back. How you finding Grock 4.5? I'm

just testing it out right now. So, we'll run it through Opus 4.8. That will test it and we'll run like six useful builds with it. So, you can see what we'll test out with it first. These actually, sorry, nine games that we'll test out. See how it performs. Once we build those out, we'll test it more. Um, the last Grock build was pretty good, so I'm interested to see what Grock 4.5 can do as well. Let's

see how it goes. I'm feeling optimistic so far. Have you tested out DG Hey. Uh, you can get it inside the air profit boardroom link in the comments description or just go to the air profit boardroom if you want the agent OS and then you can find all the details right there. Awesome. Yeah, just some basic Grock build stuff right now, but no benchmark style testing. Seems great for current use case. Yeah, I'm I'm interested to

try it. See how it goes. already getting stuff back actually. It looks interesting. Rendering clean. No errors so far. It's looking good. Hey hey hey. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat up Heat. Heat. Heat. Hey. Hey. Hey. Oh, this is looking pretty good. It's going to be interesting to test this. You make sure you update Goldie Bench. Heat. Heat. Heat. Heat. Heat.

Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Let's see what we got here. Your AI news and YouTube videos. what got me to get a better understanding of AI and its capabilities. It got me developing my own software. Wow, congrats. That is actually amazing, Methodica. Thank you so much. That's amazing. Love it. Congrats on the journey. Thanks for being so positive as well. What we got here,

Trey? All I'm doing is watching you work out. Hey, let's keep it positive, man. We're testing out Gron Claude here. We're having a good time. Absolutely love your channel. Enjoy the content you create. Thanks so much for the positivity, Methodica. You legend. Thank you. Trey says, "I am going back and forth on a project passing information between GPT 5.5 and Grock 4.5. Grock is holding its own in my opinion. I'm seeing the same thing actually when

we're testing out from what I've seen. The results with Grock are looking really good. So once we fully tested it, I'll show you what it looks like. Uh, let's see if it's been added to Goldie Bench now and we can check it out. Yeah. Just going to ask if we've deployed it yet. deploying it right now. Be interesting to see how they perform. It's scoring an average of 7.6 which is up there with Fable 5 to

be fair. I think that's outperforming or right up there with Opus. And then we'll build out the other 45 projects as well in a second. See how it does. Heat. Heat. actually might be one one of the best. Wow. I'll show you this in a second. I'm excited to show this. Uh Grock is looking very good. That's what I can tell you right now. Wow. Hey, hey hey. back in a sec. If you want to see

what we've done so far, check it out on Goldie Bench under Grock 4.5. It's looking good. It's looking very good so far. Um, way way way way better than HY3. By the way peeps this one was pretty bad. This is looking good. This is very impressive. You can check it all out at Goldie Bench, by the way, peeps, if you've never never seen that before. It's looking good. I'll be back in a sec. Grock 4.5 has

dropped. You can try it for free and it is pretty impressive. Let me show you what I've built with it so far. We've tested it on eight different builds. It's very impressive. You can see our results so far on Goldie Bench if you want to have a look. But for example, if we check out this game right here, you see right here, like the the graphics quite nice. It's fun to use. I mean, the the camera

angle could be a little bit better, but overall pretty cool. Let's have a look at next. So, this is kind of like a openw world RPG game. As you can see, camera angle is nice. Graphics feel good. Good physics in there. Let's have a look in the sky. Try and find the dragon. There it is. We got the mountains over here. 3D looking pretty cool. We have much better outputs than when I tested something like, for

example, HY3 from 10centent. It's looking really good. Let's test the next one out. So, this is a racing game now. That is fun. That is pretty cool. Uh, definitely works. Decent graphics. Let's try it again. Little bit hard to control, but I guess that's the point of the game. Let's try it out here. Yeah, it's pretty good. Pretty good. Um, a little bit buggy in parts, but overall nice, fun to play etc. Let's check the next

one out. So, this is like a open world RPG game and very easy to control. It's actually got enemies as well, which is good. very big expansive open world style. Yeah, it's looking great so far. Let's try the next one. So, this kind of voxil craft game kind of like Minecraft as you can see here. And it just built this out in like one or two clicks. I mean, literally most of these are just one prompts.

It's very very smooth. Test super nice. If you actually compare this, and we will compare this versus, for example, Opus 4.8. It'll be interesting to see where it goes. I think this is the only one that didn't look very good at all. So, you can see this is whatever it's supposed to be is terrible. So, I'd say that's probably on the worst ones. And then we have this game, which is super fun to play. Works really

well. Not too much detail. I have noticed that in some of the the builds it's like a little bit hard to control some of the games. For example, the controls are opposite sometimes, but overall fun to play, works really nicely, easy to use, and bear in mind like I'm I'm already subscribed to Twitter, which means like if you're using something like this, you don't need to pay any extra. You get it included, which is pretty amazing

in itself. So, eight oh, eight tests, seven good builds. This one was not very good, but the rest were actually pretty decent. How does it compare versus other models? Let's find out. So, if we filter on Goldiebench just to Gro 4.5. Let's compare the builds. I mean, this isn't really a fair comparison, but if we wanted to compare, how does it do? So, we've got Fable 5 versus Grock 4.5 just to show you the difference. So,

This is the racing game we talked about before. Then we have this from Fable 5. And you can see there is a clear difference in the quality, the graphics, the control, the details, even like the UI. It's much nicer with Fable 5. Now let's have a look versus Opus 4.8. So actually on Opus 4.8 this generation didn't work at all. Let's have a look on Gro 4.5. So that actually works. Pretty nice, easy to use, fun

game, etc. Um graphics are much nicer. Does feel a little bit like Lego, but apart from that it's good. Let's have a look at the next generation. So, this is Opus 4.8. Really weird. Sometimes you create these games and like the controls are backwards, if that makes sense. So, like what you think is walking forward is actually walking backwards. Uh, apart from that, not bad. Like, I like the the effects. Everything else, let's have a look

at this one. I mean, this one is not far off. I wouldn't say there's a huge difference between them. If anything, this feels a bit smoother and less laggier. Let's compare the next one. I would definitely say Opus 4.8. Well, actually, yeah, I would say Opus 4.81 on this one. It just looks a lot nicer, better UI, etc. The only thing is you'll see that it's a little bit laggy when we're playing this. If we compare

that to Grock here, the gameplay itself is much smoother, but the details in the UI not as nice and it's a little bit a little bit buggy in places. Definitely lost on this game. Let's have a look at the racing game. Oh, what happened here? This is Opus 4.8. This is Grog build. So, I mean, you know, on some tests it's up there. I think Elon Musk actually said it's like Opus class, Opus level, which I

would definitely agree with. I think like, you know, Grock 4.5, it can create some amazing stuff. Nice generations, nice UI, works pretty smoothly, outperforms Opus 4.8 in some places, and also when you're, for example, building something, it just feels smoother in general with Grock 4.5 versus Opus 4.8. So, here's one more. So this is Voxilcraft. This is Opus 4.8. Bit of a disaster on that one. Let's have a look and compare it to Grock build here.

I mean this is much better, much nicer, much more interesting, etc. Right? And actually works. So there is a lot of instances where Grock 4.8 is outperforming it. Now Grock 4, not uh 4.5, not Grock 4.8, uh Grock 4.5 just got dropped. I've been testing it for about an hour now. So once we have more time to test it out, I will create all 47 tasks directly with Grock 4.5. They're all generating in the background right

now. So it'll be interesting to see how they perform side by side. But overall looking very promising. This is by far the best version of Grock that I've seen so far for coding, for building stuff out, for getting graphics. Definitely the best. So you can feel an instant difference, especially when you compare it side by side. So let's see how it performs on the benchmarks. Again, who's a lot of people ask me this, which benchmarks do

you trust? Do you trust benchmarks? Which ones do you like? I always say test it for yourself. And also, if you want to see what I've generated with it, you can check it out on Goldiebench. So real world engineering excellence. Grock 4.5 was trained on data sets, spanning knowledge in coding, science, engineering, and maths, um, good intelligence, etc. You can see how it performs. So this is Gro 4.5. Not quite as high on Deep SE versus

Fable and GP 5.5, but at the same time it's outperforming Opus 4.8 Max by quite a long way. Again, test these yourself. Try it out yourself. See what you think. But for me personally, I would say that Opus 4.8 is getting beaten in some tests by Grock 4.5. And you've seen that side by side today as well. So I would try it out yourself. Now you can use this with Hermes as well. So if you want

to use Hermes directly with Grock 4.5, you can create an oorthth login with it. So you can see we've got Grock 4.5 plugged into Hermes inside our agent OS over here. We've added that as a new profile and then it actually works pretty quick to reply. So we just said like yes, you working? Yeah, it is. Yeah. Yeah. Blah blah blah. If you want to change your model inside Hermes to Grock 4.5, you can just go

to your terminal here and then just type in model and then from there you can select the model that you want to use. So you would scroll down to XAI gro a and then just use it from there. That's how you can get access to it. So it works inside Hermes. It performs pretty well. It's right up there with Grog. We also tested it inside our Grock CLI as well. So that's plugged into this system. Again,

we're using Grock build CLI on X. You can see the stuff that we built inside our workspace here. So the cool thing about that is we could kind of use it as a standalone CLI here. We can go back and forth with it. We can ask it to build stuff and then inside our workspace will actually get the generations back. Also, another thing that we built out with Grock 4.5 is the studio system here. So we

can go inside Hermes studio. This is inside my agent OS in the airoft boardroom. You can go inside Hermes Studio. You can ask it to generate an image and then it will give you an image or a video or voice note back. So let's try this out. So we've asked it to generate an image of a cat and very very fast, right? That literally took about 5 seconds to generate an image. Let's try now a video.

So, we're going to say create a video of a cat eating cheese and grow can create some pretty nice videos as you can see right here. So if we look at some of this, the way that I've previously used it is we actually built it into our video agent over here. And then what it can actually do is we can type in a prompt into the video agent and it will do the research directly or whatever

topic we type in. We can change the length over here and then we can actually use Grock to generate the B-roll with an AI avatar. So if you want to see an example of that in action, this is an example. This is a custom workflow we built into the Asian OS. So, you can see that it's added the B-roll inside a video with the AI avatar, done the research, created the voice over, plugged it all together,

and we create an awesome video like this. As you can see, another powerful example of how to use Grock 4.5. A lot of people be like, "What's the use case? What's the use case or how to use this stuff?" Or, you know, Grock 4.5 isn't any good, etc. But you can see it clearly is good. It can create really amazing stuff, but the most powerful way to use it by far is to create your own custom

workflows. So to create a video agent that can combine the power of Grock 4.5 video generation and then also generate the full video with an AI avatar and a voiceover over the top. Like that's insane. That's how you use it. 99% people don't know how to do that. Now if you go into Grock uh sorry into Hermes over here, one other thing that we've actually built out is Hermes Oracle. Now with Hermes Oracle, we have Grock

plugged into this system so that it can pull the latest news inside our agent OS for whatever topic we ask it to do. So for example, we can actually say okay uh every 24 hours search for the latest trending topics in my industry and then from there create this system where it pulls in the latest news, categorizes it from Grock in terms of what's the most trending. So you can see this scores 92 versus this scores

65. So obviously meta is not as big news as Claude Fable 5, which makes sense. And then we can actually draft and publish straight to our WordPress blog the SEO content. We can draft the social media content and we can see the original tweet over here. Right? That's the power of this stuff. But you want to have like custom workflows. And that's the way that I would get the most out of graph 4.5. If you're wondering

how I do it personally, so let me show you an example. I will literally just go into Claude Claude desktop every I spent about 3 to four hours doing this and I would say to Claude, "Listen, we want to build this studio inside Hermes agent that combines OOTH with Hermes. Create a new profile for it. Create the Hermes Oracle that pulls in the latest trending news. And then also create a studio where we can generate images,

videos, and voice notes using Gro or Miniax as well." And so when you build out those systems, you tell Claude, Opus 4.8 or Fable 5 to build out for you, it's insane. Like there's there's so many more powerful options you get with that. And if you're wondering how to do it, you just go inside Claude. If you wondering why to do it, well, you get much smoother workflows and everything's inside one place. If you want my

systems already set up for it, you can get that inside the off boarding. But that's the whole point of this. So, it's really impressive. I mean, bear in mind like I don't have to pay any extra for this as well. Uh I again don't rely on benchmarks. I just test out myself and see what we think. If you want to get access to it, there there's three ways really I think to use Grock build which is

you either use it inside the API or you would use it inside the CLI which we've got here or you would actually plug it into Hermes agent and then you can create custom workflows like Hermes Oracle or for example the studio or for example the chat, right? Those are the best ways to use it step by step. So it's pretty powerful. Let's see how long it takes to respond. So, I'm just going to say, you know,

what did we do yesterday inside Hermes? I just want to see how long does it actually take to reply. And also, the other thing that surprised me is I haven't hit any rate limits yet, and I would expect to hit them pretty quickly when you're building out the sort of games that I showed you before. We've been testing all morning relentlessly. Um, but it's not hit any rate limits yet, which really shocks me to be fair.

Like I thought that would, you know, run out of tokens pretty quickly, but not not so far. Pretty good. You can see it's quite slow here. Now, I think that's because it's using tools, but I would say it's pretty slow inside Hermes. So, just be aware of that as well. It's like, yes, you can blog it in a wolf into Hermes. Is it fast? Not from what I can see so far seems really really All right.

I actually don't think as well when you know when I look at the demo that they've created here, it's not that interesting. Like they've really underplayed it. I don't think that's actually quite an interesting prompt. I think they could have created a much better demo there to be fair. Also, if you look at the token efficiency, this is pretty interesting. So, according to the research here, it's faster than flash models and it uses less tokens. So,

if you combine or compare Opus 4.8 Max versus Grock 4.5, this apparently uses 4.2x fewer tokens. That's pretty nice. Like, that's a big headline. Super useful. Again, in terms of pricing, I'm just using gro off, which saves a lot. But yeah, overall, super useful model. I think the best way to get the most out of it by far is to use it inside the Gro CLI and build with that directly. And then at this point as

well, you know, bear in mind like when you're using Grock, it's the only one that actually has access to real-time news as well from Twitter, which I think is super useful for automations cuz a lot of automations rely on like real time, especially if it's like around content creation or for example social media content or research like you need real time data and I think twitter and grock is is probably the best way to do that.

So overall, awesome. Definitely check it out. Best way to use it inside the agent OS because you're going to combine it with custom workflows and get way more out of it. So for example, the video agent that I showed you earlier, the Hermes Oracle system that pulls in the latest news, the studio that can generate images, videos, and voice on the spot. Like this stuff is super powerful. That's how you get the most out of it.

And also bear in mind like you can use a memory galaxy with it. So we've got a memory system here. It has all my context. It takes notes on us. It's being updated by all my agents. So every time I use it, my notes get plugged into automatically into the system as well. So for example, we've added Hermes Asteros directly into our memory galaxy. So like every time one of my workflows with Hermes runs, it automatically

updates my memory, which my agents can then use for context in the future. So it's a self-improving loop because the memory gets better every single day. So if you want to get my full system for all of this, feel free to get it inside the AI profit boardroom link in the comments in description or go to the profitboard.com. This is my automation community that helps you save time, grow and learn with AI automation. Inside the community,

you can ask questions, get help and support in real time. Loads of great people inside there. I personally answer the questions inside here daily, too. Inside the calendar, you can jump up weekly coaching calls, ask questions, share your screen, meet other call members. inside the classroom, you can get access to all of my best training. So, if you're a complete beginner, you can get the beginner to expert course over here. And if you want our agent

OS system, you can get that over here. And we update this daily with a zip file that you can install a video tutorial. And also every time something useful comes out like for example we createdization playbook we create a full system for it right here as you can see with a video tutorial and an actual step-by-step guide. So feel free to get that link in the comments description or just go to the proferform.com and then also

inside the map you can meet people locally who are building with AI agents like you. So feel free to get that and obviously inside there plus you can DM me anytime too. Cheers watching. What's up? I am using Grock build now and it's impressive. It feels a little less intelligent, but it is faster and cheaper. There we go. I just want to know how you create so many how how you create so much content. So, if

you want to learn more about that, we have inside the AI profit boardroom, we have a full system on YouTube where I show my whole workflows here. So, if you want to learn more about that, you can get my full sixeek course on how to create content over here. And that will help you a lot, I think, if you're wondering how we do that. I get that question quite a lot, but yeah, you can get the

full training on the air profum if you want to learn more about that. How would you define vibe coding? I mean, if you're asking that question, what are we doing here? I I would say like it's just like creating coding something out. Just sort of jazzing around with AI coding tools. That's what I would say it is. How would you define it? Um, let's have a look. We got this getting done over here. Oh, what is

going on with Opusly letting us down here? Today we're going to be answering questions about the agent operating system. How to use it, how to get the most out of it. This is a powerful system where we combine a memory system. We take the latest updates. So, for example, like Grock 4.5 just dropped. We've already created a Grock CLI inside the system where we've tested out loads of different core builds with this and it looks really

really impressive. And bear in mind, this is free if you're already subscribed to Twitter, which is awesome. We've got, for example, like free clawed code plugged into HY3 and Omniroot that just dropped over the last 7 days. And then, for example, with the new Gro 4.5 update, we've got Hermes Oracle plugged into it. We've got Hermes Astros. We got a voice powered AI agent. This is by far the most powerful way to use AI agents. 99%

of people don't know about how to use this or that it's even possible. And we've already created all these amazing systems around it. So, we actually give this to AI Profit Boarding members and then we get a lot of questions daily about how to use these systems. So, what I'm going to do today is just answer those latest questions. Let's get into it. So, we got a question here from Jean. And Jean was saying, you know,

with Ruffalo, is it still worth using or is open claw or Hermes better? And I would say Ruffalo was great grow. If you're not sure what Ruffalo is and you're watching this, by the way, Ruffalo is a system where you can have like multiple sub aents working with whatever agent you're using. So for example with Claude code you could create like a full team of agents. I would say that Claude code actually replaced it with Ultra

Code. So they created their own version of Ruffalo anyway. So you don't need it. And then also if you go into Hermes Hermes automatically has teams of sub agents as one of their latest updates as well. On top of that if you actually go inside the camb board of Hermes agent you can visualize the sub agency you create. So if you want teams of agents, you don't need Raf anymore because you can either do that with

Claude and Claude Ultra Code or with Hermes. But either way, you don't need Ruffalo. It was just really good a couple of months ago when none of those features were out for either Ruffalo or Hermes. So, we got a question here from Andrew, and Andrew was asking like how to set up Open SEO. So, there's a full set of instructions on the GitHub right here. As you can see, this is a powerful system for rank tracking,

for automating SEO, for having different SEO workflows, and then you can plug it into your agent OS. So for example, we have this full SEO section for SEO inside the agent OS here. And then we have open SEO there. But if you're still struggling to set it up, what I did, and this is what worked for me personally, cuz I never set up Docker myself, is just get the details of the GitHub, paste that into Claude,

make sure you're using either Fable 5 or Opus 4.8, and then get it set up for you. And then just keep it looping until it finally works. John is asking has anyone attached Hermes to Abacus? So I think Abacus actually have a cloud hosted version of Hermes and you can use that directly with Abacus but personally I would just use it either self-hosted or in a VBS and actually Hermes have just released Hermes cloud today where

you can actually host Hermes in the cloud and use it directly there instead. So, I don't see the need for using it with Abacus unless you want all of their tools. But when I tested out myself, it was pretty slow. And I think there's better ways to use it when you self-host it yourself and customize it as you want. So, we got a question here. Uh Nick was like, you know, with the boom of AI and

marketing becoming more saturated and everyone having an AI agency, which I actually don't know many people who have a big AI agency, like there's a few people implement it, but there's not that many people who implement it consistently and sell it to clients. So, I still think it's wide open right now for most people, but I haven't even seen any good white label options. But Nick has said, "Why not make the business yourself and implement it

for yourself?" Yeah. So, we actually implement it in every single one of our workflows inside our agency. So, for example, my whole team uses it and we've actually scaled back the team a lot because of the automations that we have. So, I think if you already have a business, the best way to start with AI is not to create a new agency, but to implement it inside the existing workflows you have, inside the existing business that

you have. So Jason was asking about how to set up obsidian. Should you use it with a separate human vault and a vault for the AI or should you just have one main obsidian setup? So for me personally, I just have one obsidian setup. I don't have like multiple setups. I think it's better to have it all in one place just to keep it simpler. And also when you're setting up Obsidian, the more complicated you make

it, the more tokens it's going to use and the more difficult it's going to be to implement it. Whereas, if you have it all inside one place where you check it and your agents check it, then it's a lot simpler. It saves time and it uses less tokens. Jose was asking about carousels with NA10. One thing that I'd recommend, this is the way that I would start with NA10 workflows in the future, is just ask NA10

if it can be consistent via the workflow builder. So, there's an AI you can use inside NA10 now where you can build the workflow via chat. And I would just say make sure that every image is in uh is consistent inside the workflow. There looks like there's a really good workflow template as well, which I'll link to below. So, we got a good question here from Hermes, uh, sorry, from Joshua, who was saying, "Hi, I've been

using Hyper Agent, but I wanted to get set up the Hermes agent with Asian OS. Seems pretty similar to what I'm currently used to with the ability to use open source models and lesson AI usage. Where do I go to learn about how to do this?" So you can get the full agent OS system here with the Hermes setup inside this section. So you go to the classroom inside the AI profit boardroom and then you go

to new daily updates and you can find the zip file to install it a full step-by-step guide in the video tutorial here for setting up a VPS. I'm going to show you two different examples. So I think that Paul actually got it set up this way instead, right? So you can actually get the agent OS multiple devices or you can actually use something like Hoinger, I believe. Yeah. So, a lot of people using Hostinger VPS that

seems to be pretty popular. Jing was asking how to upload contacts into go highle. So we use high level personally as well. There's a really good tutorial for this which I'll link to below. It's quite easy to do with a CSV. We got a question from Elsa. What's the best practice to set up agent OS safely? Should you set up the agent OS on a completely different machine with his own Apple ID or on a VPS?

Uh, wants it to be safe but also functional. What's the best setup? For me personally, I just have it running locally. You can also ask the agent when you set it up to reduce the amount of permissions it has so that it has to ask you for approval for everything if you want to play it really safe. I'd also recommend that you don't do anything you're not comfortable with. So if you're not sure, always stray on

the side of not doing it right. Like if you're not sure about setting something up or you're not sure about using a certain skill or CLI, then just don't do it. Additionally, I wouldn't have it on a VPS. I think if you look at all the problems that came with Open Claw, there are a lot of problems with it. And like I say, for me personally, I just have it on one machine. You if you want

it to be really, really safe, then you can actually set it up with local only models and have it running offline on a separate machine without access to the internet. That's another option if you want to play it really safe. So, it depends on what you want it to do. And bear in mind, like the more functional you make it, the more powerful it becomes and the harder it becomes to control. So you want to make

sure that if you do want it to be functional that you're comfortable with the way you set up We have a question here from Stephen who says, "What are you doing to build your second brain?" So, as I showed before, we have the memory galaxy over here. That is my second brain. Super powerful stuff. Gets updated automatically every like every time we use our AI agents inside the system. So, it all links together, which is beautiful.

And then you can see here that Steven says, "What are you doing to build it? I have tons of conversations across Codeex, Chachi, Clone, and Hermes, but I'd like them all to use a shared knowledge base. So, what would I recommend? I would recommend that you use Obsidian and then have an agent OS that brings it all together. But you can just start with Obsidian to keep it simple. And we have something called the infinite context

engine which is a system that combines OMI that takes notes for me that plugs it into Obsidian and then Obsidian plugs it into my agent OS system and that works really well. So if you're watching this, you're like, what is OMI? OMI is a an open source project that can listen to my microphone, watch my screen all day, and then from there, it can actually take thousands of memories as you can see. And then after that,

what it's going to do is we can actually export the newest memories to Obsidian, right? So, for example, we can sync that to Obsidian as you can see here. Now, once that's done, it's then going to export the information over to Obsidian. And Obsidian looks something like this. And you'll see a full knowledge graph here. Now, once we've done that, we actually add it into our memory galaxy inside the agent OS. So, it all syncs together.

And if you want a really detailed version of that, check out the infinite context engine training that we created. Really powerful stuff. So we got a question here from Pedro which is right now what type of AI automations are most in demand and should you use NA10 Claude code or Hermes instead? There's so many things that he finds himself getting lost in what he should build and try to sell. So from what I see the most

popular service that people want is an agent OS system because clients know they need a dashboard like this but they don't know how to implement it. So if you can build that with claw code or you can for example use our system customize it for clients and then sell it to them that would work really well. And I think out of everything, this will be the most powerful to sell because clients want more leads, they want

more content, they want more customers, and the agent OS system can be customized for that whole system. Plus, it's a way to create almost like a SAS tool that's fully customized without having to be a developer. So for example, if we have a look at the systems that we've set up, like every problem that I had, I've now set up inside the agent OS. So for example, I used to have to spend a lot of time

creating videos, so we created the video agent. I used to have to spend a lot of time creating SEO, so we created the SEO agent. I used to, for example, struggle to have context across my agents, so I created the memory system. And then I've shared that with the whole community inside the air profitable boardroom and that's something the clients would be interested in too. So you know if you look at it from the client's perspective

it's like right okay what do they need? Well they need more leads they want to grow. So how do we get them more leads? Well we can create more content for them using the video agent or the SEO agent. We can generate more leads for them using the outreach tool over here. We can help them organize all of their data using the memory galaxy. And that's the most powerful way to help clients. I think that's what

they're really interested in. and GPT 5.6 is about to drop on Thursday. Can't wait for that. I've already tried out Grock 4.5 today. That's really good as well. So, that just dropped, too. As soon as GT5.6 drops, I'll be testing it and adding it inside Goldie Bench as well. So, I can't wait to see how it performs. A lot of people saying it's not quite as good as Fable 5, but let's see. for the people who

got early access to it. screen. So Chad was saying the his Hermes agent is using a lot of tokens. So for me personally, if I'm using something like GLM 512 with Hermes, I don't use the API. I would just use the coding plan. And then that way you've got a flat cost for the coding plan. You don't seem to ever run out of tokens with Hermes and therefore there's no extra costs. So, it's limited to whatever

the coding plan you subscribe to costs. Also, we have a token minimization playbook. So, if you check out the token minimization training inside the classroom that shows you multiple different ways, like 10 different ways to reduce the amount of tokens your agents use, which solves a massive problem too. So Garrett has a few questions here about the best setup for paperclip and Hermes. So what is the best way to use paperclipip together with Hermes agent for

implementing skills and tasks and everything else? So the way that we set this up inside our agent OS system is we created an AI SEO agency with a bunch of agents that work together in a team and that seems to work pretty nicely. No, I don't use it that much. And if you're wondering what the best use cases are, it may be a case that you don't actually need it cuz usually the use cases stand out

to you very obvious. So it might be the case you don't even need it and you're over complicating it. But if you do want to try it, what we actually did is we got it to work on a bunch of different tasks together. Like for example, creating tools, creating websites, creating relevant mini apps related to our SEO agency and then it would just create and work on those for like hours autonomously as a team. But honestly,

I think if you're not sure how to use it and you're wondering what the use case is, it may be a case you don't actually need it and just one hermitation would do the job. So, for example, if you were looking at paperclip and you're like, "Wow, okay, well, you know, I really want to create a team of designers. That's something that's going to help me a lot. It's going to save me a lot of time.

Okay, let's go with that." But if it's a case that you feel like, I don't even know what the use case is, then probably in that instance, you don't even need it. It's unnecessary. You're just making things too complicated for yourself. and that's basically it. We've answered every single question inside the AI profit board today. I do this daily for all the members. So, if you want to ask me a question, feel free to join and

post inside the community, inside the classroom. You get access to all of my best trainings, systems, lessons, etc. So, if you're a complete beginner, you can go from beginner to expert over here. If you want to get my agent OS system, you can get that here. And we have a last update date. You can see the full video tutorial step-by-step guide and the zip file down here. Also, we add new trainings based on what's actually useful.

So, you can see, for example, we've just added a new token minimization playbook to reduce the amount of tokens you use so that you need less resources to run your agents. We also, for example, created a new training on HY3 from Tencent and we're building new stuff out all the time based on what's actually useful and can help you. You can also direct message me if you have any questions personally. And inside the calendar, you can

jump on week coaching calls, get help in real time, ask questions, etc. Inside the map, you can meet people in your local area who are building with AI agents and using the agent OS. And that's all inside the air boardroom. Now, you might also say, okay, well, can anyone use an agent OS system? Can anyone build it? So, let me show you some examples here. Like, we've got so many testimonials and wins. So, for example, Jacob

set up the agent OS in about 10 minutes and we've got 196 pages of testimonials and wins just like this. So, you can see here, Rick joined 30 minutes and already had the OS set up using claw code. Like there's so many powerful s uh so many great people inside the community who are non-technical can set up an agent operating system are getting great results are sharing their dashboard automating and implementing and customizing everything too. So

if they can do it and I can do it then you can do it too right like anyone can use an agent OS and build with it and it's a really powerful system once you understand it because you can use it for your business you can customize it how you want you can save a lot of time and also potentially you could sell this to clients as well. So thanks so much for watching I'll see you

on the next one. Cheers. Let's see what questions we got here. I am vibe coding too. Nice. You are doing a great job. Keep going. Thank you very much. Never seen such a dedicated person. Hope you have a lot of success. Thank you very much, sir. That's super kind. What's on sw.7? Is this Oh, this is the new benchmark, right? I assume. If it is, the reason the whole reason that I created the Goldie Bench system

is because the benchmarks I saw were just crazy. Like it none of it made sense to me. Uh and you know, I I saw that with, for example, a lot of models were coming out, especially local models where you see them on the benchmarks and then you test them and you're like, how how is that even possible? How is that even relevant? And so we created the whole goldie bench. Uh I like to look at S

sw and stuff like that and just see what it is. But in general, do I take it seriously? No. I think the best thing you can do is just test it out yourself. Which is the most powerful agentic AI system you have seen until now? Definitely my agent OS system. I've not seen anyone else build anything as good or as powerful as this? I mean like for example, Grock 4.5 drops, we've already got it built into

the system. Uh, HY3 dropped. Okay, great. Let's create a free AI coder inside the system where you can code, you can preview it, and you can see everything you've built and save it later. So, I don't know of any other Agentic system that's as powerful and as easy to use as this. And, you know, we've got 196 pages of testimonials and wins from people inside the community. So, I know that it's it's really good. It, you

know, literally I haven't seen anything better. Josia is here. Thank you very much, Josia. Thank you for for joining today. Always good to see you, mate. Do you really think GT 5.6 will be better than Fable 5? No. As I mentioned before, I think that it will be slightly behind Fable 5. I think it'll be up there, but I don't think it would be as good side by side. So, yeah, personally for me, and from what

I heard from the beta testers as well, they all said like it's good, but it's not as good as Fable 5. So, that's my honest expectation. is like it'll be okay, but it won't be as good as Fable 5 and everyone will just keep using Claude like they have been for the last year or so. Um, yeah, I think like there's a reason Claude is so popular and so powerful. Let's see how we're getting on with

these tests. So, we got that over here. Let's have a look at Hermes Cloud. Do you know what? Like Opus 4.5. What is going on here? Let's see if we can get it working again. How much does the agent OS cost to run? So, you can actually use it for free because there's a local model section here. There's Omniroot, which is a free way of using over 90 free APIs with fallbacks as well. So, you never

get rate limited. And we also have HY3 coder inside there, which is free. Obsidian is inside there which is free. So if you want to use it completely for free, you can. It's up to you. Like you you can put as many resources into it as you want, but if you don't want to and you just want to use free models, you can too. Did you hear about GPT6 and Fable 5.1? I didn't hear that. And

I say I I only pay attention to relevant sources. So, I can't see Fable 5.1 coming out anytime soon, but if it does, happy days. I'll be celebrating with you. The whole point of the agent OS is to build the most powerful system that helps you grow your business to generate more leads to save time and to scale. That's the whole point of the agent OS. What's on perplexity computer? Don't pay attention to it at all

anymore. Honestly, I think like the if you look at the agent OS like like it's way because you can build your own computer, your own agentic operating system and you can customize it as much as you want and it can be way way more powerful like I mean what other system can you look at the Hermes Oracle like even for example if you look at the Twitter explore page it doesn't give you as much information and

it's not as useful as this if you, for example, wanted to monitor your competitors. I don't know of any other system that's as good as this for monitoring competitors. So, like everything I've seen out there, I've tried to make this even better inside the agent OS and way more powerful. And I think like it's way more useful to use this than any other system I've used out there. So, I think for example, something like computer just

not never going to use it. Probably will never use it again simply because the agent OS is way more powerful. I don't I don't see the point of using like a mediocre tool or getting distracted with mediocre tools when I can just build my own better system like you can see right here. And you know, publicity computer is probably okay for people who are not aware of like how to build an agent OS and it it

probably doesn't do a bad job, but it's like why would I be limited by someone else's system when I could build and own my own, customize it as much as you want and then improve it every day. Like that's way more powerful because then you don't have to rely on uh another company. You own the system and you improve every day. Whereas for example to make an improvement on computer you can't right. Someone else owns it.

But best uh yeah I mean like out of everything the best coding tool is claude for everything. Literally for everything uh the best is Claude. So, you know, whatever you want to build, I would point directly to Claude. I don't think Codeex even comes close at this point. Uh, also the other thing I found with Codex is like it just runs out of tokens really quickly. Do you think Grock has a chance of really pulling ahead

soon? I don't see how it can because for example if you look at Grock 4.5 like it's barely competing with Opus 4.8 like it's still behind Opus 4.8 so I don't see how it can overcome or or overtake something like Fable 5 which is way more powerful like after you've used Fable 5 when you go back to Opus 4.8 like what are we doing here? So, I don't see how in the future Grock is going to

overtake them unless something crazy happens even with all the stuff that they have. Like they've they've always been be behind, you know, ever since like last year when the first model the first series models came out. So, I just don't see how they're going to overtake someone like Claude. Claude are the best in the world at what they do, you know? That that's undeniable. The other thing that I would say is like if you look at

Claude, they're very very focused. So they don't try and like do images and video and voice like Grock or ChachiT. They're just like, "Nope, we want to build the best coder in the world. We're going to focus on that." And I think when you're that focused, how can you beat someone like that? Right? How are you ever going to beat someone like that? So that's the way that I see it. What else we got here? Have

you used Opus models with open code? I I haven't recently. So I've used Open Code in the past, but I haven't used Opus recently. Um, so yeah, it'll be interesting to see how that goes. Uh, but I have heard that open code is a lot more token efficient, so it could be worth testing. I was tempted to build it out into the agent OS, but I just didn't know enough people. Claude is Claude. Wise words, my

friend. I mean, that sums it up perfectly. Claude is Claude. Exactly. It will probably take more time for X AI to get ahead. Yeah. And I I don't know if they I mean honestly if you look at Grock or you look at Elon's companies like and I I really respect him like respect him you know a lot like for what he's done and what he's achieved and obviously he's he's done unbelievably well but like I don't

really see him being the best in any of the fields that he's in. Uh, and maybe you disagree with me, but I just don't see like he's is he the best at anything. Whereas, for example, if you look at Claude, again, Claude, the founders of Claude, they're just focused on Claude. They're not like doing loads of crazy other projects. They're not trying to build cars and all that sort of thing. So, it's like, well, how can

you out compete someone who's that focused at one thing? I don't I don't see it in the future. Yeah, I see what you mean, but I still just don't see Claude overtaking uh being overtaken by Gro Opus 4.8. Uh, of course. Yeah, Fable 5 is Fable 5 crushes Opus 4.8 like by a long long way. If you actually look at Goldie Bench and you see our tests, we tested on like 47 different projects and Fable 5

was just unbelievable. Sometimes Opus gives better results than Fable 5. Well, sometimes Fable 5 switches to Opus, so you got to bear that in mind as well. But yeah, if you I mean this is only from my own experience. So if you want to check out Goldie Bench, have a look for yourself, see what we've built, compare the test side by side, and then see for yourself what you think. Yeah, I would agree. I would agree.

I think they're miles ahead of everyone when it comes to coding. Lord, way better than everyone. Oh, we got all 48 out of 50 builds from Grock. Let's have a look what else we got here. Gro 4.5. I don't know if it's been deployed yet. Let's have a look. Oh, yeah. We got all the demos here. Wow. Let's see what we got. So, this is a video that it's created. using Remotion. It's looking bad. It's not

looking bad. It's pretty clean. So, it's I mean, it just created this video in the background. It's pretty awesome. Let's have a look what else we got here. The parachute test. Boom. Shakalaka. Oh, it's okay. It's okay, but it's not as good as I mean, look at the details here. Not that good, is it? Let's see what else we got. Oh, look. Okay. Game's pretty cool. Yeah, we're going to have to do probably a comparison side

by side versus something like Opus 4.8. Oh, now we're talking. Look at this. Pretty good. I'm very impressed by Grog. This is by far the best version of Grog that's ever dropped by a long long long way. He is up there with Opus 4.8 for sure. Look at that. Nice. Uh feel free to post about that inside the AI profit boardroom and you'll probably get a bunch of people who are interested in trying out. Yep, they

did. Grock uh got cursor. Do you think Claude is planning on cutting down? No, I don't think so. I I I think it will always be like that. Hopefully, we get Fable 5. What do you mean by one to one level? Best deep research system. I don't really use deep research honestly, but I think out of everything that I've seen or tested, probably notebook. Like I think notebook creates the best research, the most useful research, but

I don't know anyone who really does much deep research personally. It seemed to be like one of those systems that came out and everyone thought, "Wow, that's amazing." But then never actually used it. And there's a lot of stuff like that. Grock 4.5 plus Hermes agent. How good is it? So, he's pretty good so far. I've tested out. Now, the cool thing is with Hermes is that you can actually get free access to Grock 4.5 if

you're already subscribed to Twitter. So, it doesn't cost me anything extra to use Grock 4.5 inside Hermes. Now, from what I've tested, I'll show you uh we've actually tested out like 45 different builds with Grock 4.5, which I can show you in a second, but from what I've seen so far, Grock 4.5 is is right up there. It's not quite as good as Opus 4.8, but it's right up there. Maybe the closest probably better than GLM

5.2 for competing with uh Opus 4.8 from Claude. So, it's a really good model. However, when one thing I've seen when I've tested out inside the chat when I build out a new profile with Hermes agent and Gro 4.5 is like it is slow for tool calling. Like it will take about 60 seconds. is Ashley said like what did we do yesterday and it took about 60 seconds or 2 minutes to reply to us inside the

chat. So we can create a separate Hermes agent profile and use it directly there. I actually wouldn't recommend it. I just think it's too slow to use. I think I'd much prefer to use something like Gan 5.2. Let's just test it out now. We'll see. Okay, like just a basic question like how are you today? Let's see how long that takes to respond with Hermes agent. You could create like something blank slate like a blank slate

profile with Hermes agent. that is an option. Um, it replies in about 10 seconds when we ask it just a basic question like how are you today? So, it seems super slow. Like I just feel if you want an agentic model and you ask a basic question, you don't want to expect to wait like 10 seconds. So, that's not the best way to use it. I'm going to show you the best way to use it instead.

What I'd recommend with Grock 4.5 is that you have it inside custom workflow. So, for example, we've buil this custom tool using Grock 4.5 inside Hermes agent that can generate images, video, and voice in like one single click. And then we can use that to basically get the power of the Gentic systems inside Hermes, but it feels like a SAS tool and it doesn't cost us anything extra. And then that way we can segment it away

from our chat. So, for example, we can use something that's actually really fast to respond like GLM 5.2 inside Homebase Agent. And then when we're actually building out stuff with it, like for example images and videos, that's when we're going to use Grock because obviously GM 5.2 cannot use uh or cannot create images for you like this, right? So for example, if we say okay, generate uh oh, we'll we'll try one of these basic ones like

you know a glowing whale in a starfield nebula for real, right? We'll hit generate here and we'll see how long that takes. So, let's try this out. We just need to make sure we switch to Grock. Let's hit generate image. And how long is that going to take, my friends? There we go. Pretty fast. And you can see the image right here. Pretty cool. Let's open it up on a new tab. Have a look what it

looks like. Yeah, it's pretty good quality, isn't it? Uh, pretty easy to generate. Takes about 5 seconds and then we have a image studio ready to go inside agent OS. Now, also the cool thing about this is I can then use that in many different ways. So, I can, for example, plug that into any of the other systems down here. Um, the other cool thing that I've noticed with Grock is that you can actually create B-roll

using Grock 4.5 and then you can describe the video that you want and use a combination of Grock 4.5 with the video agent that we built here. And then we can use uh Grock for coming up with the content, doing the research via Twitter, and then generating the B-roll for our video. And then a video agent can put it all together inside one beautiful workflow. If you're wondering how does that look in reality, well, you can

see an example here. So, we can generate a video on the spot with an AI avatar, the voice over, the B-roll plugged in generated by Grock, and also the research itself, the news itself can be researched by Grock as well because we're using that real-time API from Twitter. Pretty amazing stuff. And again, does it cost me anything extra? No. Why? Because I'm already on the Twitter plan. So, this is the amazing part of it. And I

think this is how you really get the most out of Hermes in general, but also with Grock 4.5 because you can build your own custom SAS style tools inside a system like this. They're one click away and you've got them ready to go. Right. So this is our agent OS and then we can just use these systems whenever you want. So let's say for example like right, okay, create a video about Gro 4.5 release today. So

we can type that in. We can change the length over here. So, we can have this up to like 10 minutes if you want to, but I'm just going to do a quick one just to show you a demo. So, we're going to put that 15 seconds. We can change between voice over and presenter. So, I'll go with presenter. And then we're going to switch the B-roll engine to Grock. And we're going to use this avatar

so we can see. Then we're going to say, "Okay, write the script." Now, by the way, let's have a look what questions we got here. So, I mean, hopefully we get a model That's one to one level with Fable 5. Yeah, maybe. I mean, this is a pretty good point. So, someone mentions that apparently the Oh, this is saying All right. Yeah. Yeah. So, thinking level from GT 5.6 Soul has not been tested. Yeah, that could

be hopeful, but I just don't see how Fable 5 level usage is going to be reached. Let's see. Let's see what happens in the future. Will GPT 5.6 be better than Fable 5? No, I don't think so. I feel like a lot of people are going to switch to chat GPT because of GT 5.6. Let's see. I don't know because these models change every every month. Like for me, h that is a good question. I don't

know. I don't know. Let's see how many tokens we get as well on the on the subscription. But yeah, I I can see what you mean. Like for example, if that's a really good point, like if GT5.6 comes out and then it's included inside the subscription, you can use it with your agents and you can't do that with Fable 5. Yeah, I I could see that for sure actually. You got a great point. Can you please

include Perplexi Computer inside Goldie Bench? But then it's I mean the thing is perplexity computer is not a model. It is a system. And so the problem with using publicity computer is that you would have to select the models and there's loads of different models that always getting updated inside there. So it doesn't really make sense because you're kind of using like a wrapper to test the model whereas for example perplexity computer uses Fable 5. So

you might as well test Fable 5 directly. So that's why I don't include it inside Goldie bench because perplexity computer is just a wrapper, right? It's just like a a way of having a system that puts everything together. Uh yes, you can. You can indeed. You just won't get the updates, right? But yeah, feel free to get it inside the air. Um you just won't get the updates, but yeah, you can get the latest version for

sure. You might have said this already, but what does Goldiebench actually base the scores on? I wouldn't pay too much attention to the scores, honestly. Again, I would just have a look at the actual builds. So, have a look at the builds and then see what you think from there because if you look at the builds, that's going to be way more useful in terms of these like subjective scores. even if you get the AI to

score it, it's still subjective because it's like, well, how long did you test the game out for? You know, how long did you try it out for? How useful was it? And like I don't I don't think the scores are that useful for any sort of system like this. I just think it's good to like have a look at the builds and go, okay, this is pretty amazing. Let's try that out more. Um, and I think

that's way more powerful than just uh relying on the scores which would change. Like for example the even the scale of the scores would change depending on the new models that come out because for example like once if you get an 11 out of 10 well then you have to change the whole score for everyone and that doesn't make sense. Open router fusion is also a great system. Yeah we have that inside the agent OS2. So

for example if you go to fusion over here we have a panel of models but again like it's not really a model it's a system. Um but we have the system built in inside the agent OS as well. In fact, that's the one that's leading on Goldie Goldie Bench right now. So, if you look at the leaderboard of Goldie Bench, you can see that Fusion is right at the top. We've built out 47 tasks with this.

Hermes Mixer of agents is then scoring above it. And then we've got Claude Fable 5. Yeah, that's it. I would compare the demos. I wouldn't worry too much about the uh Yeah, I wouldn't worry too much about the scores. Josia said, "I use Goldie Bench from my phone and play Dra." Nice, man. Yeah, it's a great game. It's an awesome game. We should set up a system where our AI agents can just play as well. That

would be quite funny. So you can see now it's researching and writing the script for our video about Gro 4.5 which again is a brand new update requires real-time research. Claude might be able to find that, but I think you're much better getting that information from something like Grock 4.5 because like sometimes Claude is way behind in its knowledge and then it has to do web search and it has to use more tools and it gets

slower everything else, right? This system right here will just use Gro 4.5 with the video agent, generate the B-roll, generate the research, generate the voiceover, and it's all good to go from there. I think that's way more powerful. So, we can come back to that in a second. It'll just work in the background. And then the other thing that I would say here is that for example if you look at claude which is you know obviously

opus 4.8 is like the standard sort of comparison between grock 4.5 and the best models for Hermes agent with claude you can't use your oorthth login right so if you want to use claude directly with something like Hermes you have to pay for the API but if you use grock and you're already subscribed to Twitter you can use a wall for generating images web research and also voice notes but you don't need to pay any extra

so that's a big benefit as So now we can see that the research is ready to go and basically what it's done here is it's looked at and researched the whole topic come up with the details here. It's actually use the research which is pretty useful and then also what we can do from here is just click on generate the avatar and the B-roll and that will generate the videos in the background with the B-roll create

the avatar and plug it all together. Pretty powerful. It's a system. So you got video agent that can just create and automate whenever you want. Now the final thing that I would show you is Hermes Oracle. And so what this does is it pulls in the latest news from our industry, organize it in terms of what's trending right now. Gives us an angle. We can publish an SEO blog about that particular topic in like one single

click. We can see all the content that we've published around that topic. So, for example, these are all blogs that we've created directly from the Hermes Oracle system, which basically researches the latest news every 24 hours. We consult the oracle. We get the latest trending information back and then it organized here by, you know, what's trending the most. Super powerful stuff. We can also see previous days. So, we can click between previous days and get the

history of everything, too. It's just a really, really useful system that again is customuilt. And I think that's how to get the most out of Gro 4.5. Gro 4.5 out of the box with Hermes agent not very useful. Gro 4.5 with a video agent. Hermes Oracle the studio for generating images, videos, and voice notes. You know that's literally like way way more powerful. That's 10 times more powerful. And then the final thing I would say here

as well is that you can use it to automatically update your memory system. So we have Obsidian as a memory knowledge graph plugged into the agent OS. And then inside the recent section here, we get new updates based on what's just been done with, for example, like Heromies, Apollo, and everything else. And then that is automatically logged into our memory system so that all of our agents have context about what we've worked on recently. And so

all of our agents inside the agent OS can look at the memory system, understand exactly what we've worked on even over the last 24 hours, and then they're fully trained with context and what we're doing. And I don't have to touch any of this because my agents are automatically updating the memory system for me. So that's really powerful too. You can also use it inside this system for generating images. And yeah, I think Grock 4.5 is

powerful if you have the right workflows, which is what I've showing you today. So if you want to get our full agent OS with Hermes Asteros, Hermes Oracle using Grock 4.5. We've got the studio with Grock 4.5 to generate images, videos, and voice. And then we also have the video agent that can use Grock as well to create some amazing stuff. And we can see everything that we've created as well previously over here. And we have

Grock plugged in as a CLI inside this section. And again, we can preview everything that we've created with Grock 4.5. I did actually promise that I'd show you what we've built with Grock 4.5 so far. So if we have a look at some of the outputs here, like you can create some awesome stuff as you can see, like pretty interesting, very visual style games and that sort of thing. So for example here this is a flight

simulator that we created and you can see for example we actually created a fully animated video using Grock 4.5 as well. The interesting thing about this is we just taught Grock 4.5 the skill of remotion and then it can generate a beautiful animated video. As you can see, looks super sleek, easy to create, and yeah, it does create some awesome stuff. Let me show you another example. So, this is a game that we created as well.

Pretty fun, easy to build, but yeah, I I genuinely think like Grock is a very useful model. So, thanks so much for watching. If you want to get all of this, you can get it inside the AI profit boardroom link in the comments in description or go to the AI profomroom.com. Um, this is my AI automation community for helping you save time, grow, and learn with AI automation. And if you're wondering, okay, like you know, how

what's the community like? What's the vibe like, etc. So, it's just a really positive community where we all learn and grow and share things together. So, for example, loads of people are using our agent OS system that we share inside there. Right now, we're 196 pages of testimonials and wins. You can see, for example, all these different wins and people sharing stuff, and you can see how quickly people have installed this. Like, Rick set this up

in like 30 minutes. It's a really powerful system, a great community, lots of awesome people, lots of people building amazing stuff right there. I personally answer the questions inside the community with a video tutorial every single day as well. And then also if you want to get our full agent OS system just go inside the classroom and we update it daily inside this section. So you can see the last update date you can see the video

tutorial you get full guide and then you also get the zip file to install it and then we add new daily tutorials based on what's actually working and what's useful. So feel free to get that. Also inside the calendar you can drop a weekly coaching calls, get help and support in real time. Inside the map, you can meet people in your local area who are building with AI agents like you. And that's all is available inside

the AI for boardroom. Link in the comments description or go to the room.com. Thanks for watching. What Claude plan do you have and is it worth it for you? I mean, yeah, Claude is 100% worth it for me. Whatever plan I'm on, right? I have the best plan though. I think it's a max. Uh, does the agent OS increase token usage? So we have token minimization playbooks so you don't have to worry about that. And we

also have a video tutorial on how to use free models and systems inside the agent OS as well. So if you're worried about that uh we've already fixed that problem. Don't worry. So, Grock 4.5 just dropped. Pretty powerful model. I'm going to show you exactly how to use it for AI SEO. And this is based on what's working for my websites right here, here, and here. And we've created some amazing systems using Grock 4.5 for AI

SEO that I'm going to show you exactly how to use. So let's kick it off right here. So if you're wondering, okay, how are we getting such good rankings right now? So basically, we are using the Hermes Oracle system as you can see right here. And so what we are doing with each of our websites. So for example, if we type in Hermes agent voice mode as an example into Google, you can see us ranking right

here and here as well. And then also you can see us ranking over here for that particular keyword. Right? So three times on the first page of Google. Now, how and why are we doing that? Number one, we are going after trending keywords that we can quickly rank for because they have less competition. There's a lot of search volume for them and we can easily rank for them on Google as I've just shown you a second

ago, right? With three different examples. Now, if you're wondering how do we find trending keywords, I would say there's no point going to address because that data is quite old. There's no point, for example, having a look inside your Google search console cuz these are trending new keywords. So what we do instead is we search through Twitter. We pull in the latest data from the API and we look at what's trending right now because we know

that's going to give us an indication of what people are searching on Google. And then we create content automatically for it across our websites and also our social media platforms. So you can see for example here this actually gets refreshed every 24 hours. And so essentially what we're doing here is we're pulling in the latest headlines from Twitter in terms of what's trending. So you can see for example Grock 4.5 and the new model right here.

And then we have the details of the news update, the original tweet, the angle that's recommended for us publishing content around it. And then also we can draft that content for social media or we can actually publish that content directly to WordPress in one single click. So this whole system links to my websites and then we can actually create SEO content in one single click for that particular topic. Not just that my friends because what we

can also do is we can draft content using this button here and then we can publish content across our social media to rank for that topic as well which means that we can rank with AI SEO using our social media but also using our websites and we also have a video agent inside the system that can also use this. So what I mean by this for example if we look at this video agent here is that

we can ask it to generate a video. So if we say okay create a video about Grock 4.5 and it will use Grock to generate the B-roll and the research because it will search through Twitter in real time find the latest data and news and headlines and then use that to generate a script a voice over and pull a whole video together. Now if you want to see a working example of that you can see a

bunch of examples we've generated with the video agent using my AI avatar and a voice over that's all fully edited put together beautifully. It's got a nice avatar inside here. It's got the B-roll generated Gro 4.5 and then we can rank that content directly on Google too. How do I know it can rank on Google? Because for example, if we type in this keyword, you can see us ranking with a video here and also here using

this particular method, right? So, it's super powerful, easy to set up and we can start ranking with this straight away. So, this is the system that we're using. This is what I'd recommend for you. I've got an agent operating system where all these workflows plugged into one place and then we can quickly uh gather these headlines and create content around them with videos and also SEO content. And if you want the same system for me, it's

inside the AI profit boardroom. Or if you want to get a free SEO strategy session, maybe you just want us to implement these sort of SEO systems for you, then you can get that at goldie. SEO strategy session. But either way, Grock 4.5 is really powerful. And also, you might be wondering, okay, how does this compare versus other models? This is the first time I've actually seen a model apart from GLM 5.2 compare directly compare and

compete with Grock and Opus 4.8. Right? So, let me show you an example of this. So, we got Grock 4.5 here and we can compare it versus, for example, Opus 4.8. So, if we have a look at this comparison page, by the way, these are benchmarks that we've created. We've tested 49 different tasks and if you look for example even at this particular game here. So this is created with Opus 4.8 simple basic arcade game as

you can see. Now let's have a look at the Gro 4.5 example right 3D way more powerful way more interesting better UI more fun to play and it actually out competes it. So the interesting thing about this is that Grock for the first time I've ever seen is actually holding its own against Opus 4.8. Here's another example. So this was created with Opus 4.8. Not too bad. Let's have a look. And we've got the Gro 4.5

version here. And there's not much difference between them. Like they're pretty similar. This also looks good. Also looks nice, interesting, etc. And so we've done 49 different comparisons here. And I can tell you after 49 different tasks, Opus 4.8 holds its own. Here's an example of a video we've generated. So this is with Opus 4.8 and this is with Grock 4.5 build. Now Grock 4.5, not bad at all, right? Definitely holds its own. I wouldn't say

it's better, but I would say it's just right up there. And you can see we've just consulted the Oracle here to find new information. So for example, we've got Grock 4.5 launching right here. And let's say for example, we wanted to create a piece of content around one of these particular topics. Let's see what would be good. All right. Yeah, Quen and GLM crash Asian benchmarks. If we click publish to WordPress here, that's going to generate

an SEO article in one single click across four different pages, right? Pretty interesting. And then if we look at the publish section here, we can see the articles themselves that were published for that topic. So, super easy, simple to set up. And the same for example for the video agent. Now also some people say well how do you make the content have information gain? So what we have inside this system is our memory galaxy. And with

the memory galaxy we can train our agents like Grock to generate unique content that understands me, my voice, my goals because it has all of these different memories and information about my projects. So it understands, okay, here's how we create funnels. Here's how uh I make decisions, right? It's got my decisions log. It has all of these different nodes and memories that it can understand and put together and also updates automatically. So every time, for example,

we consult the Oracle, that's actually going to update inside our memory galaxy. And then we can use Grock 4.5 to generate better content that's more unique and personalized to me because we're using this full system together. So it's pretty powerful. And then we also have inside the agent OS this research section. So we can look at our Google search console analytics here and just run research whenever we want about those particular topics. We can also go

to the SEO office. We can generate content with one keyword and one case study. We can deploy content to our website and then we can see everything that we've created in the past as well. So together it's probably the most powerful SEO system that I've seen. It's all custom made based on all of my tests recently and what I actually wanted it to do and it's really easy to use. So these are the ways and then

also you know when you're using grock you can also use the grock build CLI inside the agent OS as well. So what this does is basically you can generate like coding projects. So if you wanted to build out a website using grock like we've created a landing page here you can do that using the CLI. So you just go inside the chat tell it exactly what you want then it's going to build it and you'll see

it inside your workspace over here. So that's pretty powerful as well because then when you're actually creating the websites that you want to use for SEO it's right there. And if you want to create the SEO content for SEO, you've got it over here. And if you want to create content around trending topics, guess what? You've got it over here. And then if you want to update your memory and make sure it's all personalized with SEO

content that's not just generic fluff, well, you've got the memory system. And also, if you want to generate videos that actually rank, then you've got that system here. So everything you need to win with SEO and Grock 4.5 is inside this system. I don't know anyone else who's actually built something as powerful as this. And I think that if you're considering using AI for SEO, this is the best way to get the most out of it.

I don't know anyone else who's doing this. So if you want to get my full system from me, you can get that inside the AI profitable boardroom link in the comments description or go to the profitab.com. The agent is inside the classroom. Just go to this section over here. If you want to ask questions, I actually answer them personally with a video tutorial every single day. So every single one of these questions gets an answer from

me and I help you as much as I can. Inside the classroom, you can get all of our best trainings. We have a full AI SEO automation section over here. And then also we have new daily updates including new updates on agent OS and everything else. Inside the calendar, you can drop a week coaching calls inside the map. You can meet people in your local area who are building with AI agents like you. And that's all

available inside here. Also, if you want that free SEO strategy session, then you can get that at goldie. Agency link in the comments description and you can book in a free SEO strategy session. On that call, my team of world-class experts are going to look at your website. They're going to analyze your rankings. They're going to see exactly where you can improve and then give you advice that's personalized to you to make sure that you can

actually rank and get traffic to your website with SEO based on what's working for us. So feel free to book that in link in the comments description or go to the airpuff.com. Thanks for watching. Uh I I would say the worth and the value of something depends on what you get back. So you want to look at what are you getting back personally and then analyze it from there. Uh what you don't want to do is

ask me that question because like the what I get back from it might be different from what you get back from it. So I would just look at your own workflows and go okay does this make sense? Is it save me time? But for me, 100% Claude is 100% worth it. So, we've been testing out Grock 4.5 versus Opus 4.8, and I want to show you what we've seen today and how it performs and whether Grock

4.5 actually beats Claude Opus 4.8. I've actually been very impressed by Grock 4.5. So far, it's performing very well. It's created some really interesting things. So, let's see how it performs side by side. We've actually got the compare section over here. We can compare both of them. We've tested 48 different models side by side and tasks. So let's see how they perform. We got Grock 4.5 on the left hand side, Opus 4.8 on the right hand

side. On the benchmarks, it actually looks like Grock 4.5 is on a similar level. But what does it look like in reality? Because I don't like to pay too much attention to benchmarks. I think we've all been in a situation where things don't feel or look the same as the benchmarks that we see. And that's why I actually created Goldie Bench. So, first thing that we're going to test is the Crypt game. So, this is Grock

4.5. As you can see, pretty nice, easy to play, nice graphics, 3D works smoothly, gameplay is good, and it's created like this kind of maze game as you can see that's fun and interesting to play. Right now, if we actually have a look at the version from Opus 4.8, it didn't generate. It totally failed at the task. It created the entry page here, but nothing else. So, so far Grock 4.81 uh Gro 4.5 sorry it gets

confusing between the numbers to be honest with you. Next up, we have Dragon Realm. So, I'm going to play these side by side. This is Grock 4.5. Not too bad. Pretty smooth to play. Looks quite nice. And then let's compare this versus Opus 4.8. Now, one of the things that I've noticed straight away with Opus 4.8 is that the controls are kind of res reversed. Uh, but I would actually say like the 3D landscape is probably

better with Opus 4.8. Okay. So, I'm going to put that as a tie cuz the controls are easier with Grock 4.5, but open 4.8 wins on the game play. So, it's a tie on that one. So, so far Gro 4.5 is 1-1 and Claude hasn't won any. Next up, we have the racing game. So, this is Grog 4.5 and this is Opus 4.8. Opus 4.8 It totally failed on the racing game. So, you can see it's

like super buggy. Let's compare it versus Grock 4.5. Pretty nice, smooth, fun to play, difficult to play, but it's actually like a playable and useful game. You know, the graphics are even nice as well. So, that's pretty cool. So, two to opus 4.5, two to Grock 4.5, and zero to Opus 4.8. Let's compare the next one, which is the Skyrim style game. So, think we're going to have to go full screen to play this properly. There

we go. It's smooth. It's fun. The graphics are a little bit basic, but overall it's pretty smooth and fun to play. Actually playable. Actually useful. Very open world. Like seems like a massive game when you actually walk around in it. Now, if we compare that to Opus 4.8, Okay, I would say the graphics have more detail just on a slightly better level, but there's nothing really going on here, right? There's no, for example, enemies, there's no

characters. It's just you first person walking around. So, there's not that much interesting stuff going on. So, again, I would say the Grog 4.51. And by the way, like don't listen to me. Don't trust anyone else. Just test this stuff out yourself. But if you want to see what I've built, you can check out Goldie Bench. But I always think it's better to check the stuff out yourself. See what you think. Go from there. Now, this

is Twilight Veil with Grock 4.5. Camera angle is a little bit hard to control, but overall pretty cool. It does feel like a little bit small, but maybe that's just because we need to walk around more. Let's compare that versus Opus 4.8. Opus 4.8 doesn't get past the beginner's page here. So, you can see we're clicking to begin, but it doesn't work. So, four to Grock 4.5 and zero to opus 4.8. Now, let's have a look

at Voxilcraft. So, this is kind of like a Minecraft style game. As you can see, we can walk around, we can build stuff. It actually has a functionality build as you can see right here. And then we can select the blocks that we want as well. So we can switch between them, which is nice and useful, too. I think it's a little bit buggy over here. Oh, we got all these crazy characters running after us now.

Difficult. Difficult. All right, that was pretty nice. Let's have a look at Opus 4.8. Pretty bad. Pretty bad. This is super basic. So five to grow 4.5 and zero to opus 4.8 so far. Now we've got this one here as well which like a virtual runner. It doesn't work on opus 4.8 as you can see here. And it doesn't work really on on gro 4.5. I wouldn't say either of them actually created an infinent here. This

is super laggy and buggy and you just can't play it. This one you can't play at all either. So I would say they drew on that one too. Then we have the driving game. So this is Grog 4.5. It's it's a bit buggy in places. For example, like we can't drive through there uh sometimes. And then there's not much detail. You see how it's just buggy? It's like driving into nothing. Whereas, for example, if we look

at the Opus 4.8 version here, it's pretty nice. It's a little bit laggy in gameplay, but overall pretty nice. So I would say that Claude won that one. Then we have the edited video. Right. So this is the video generated using Remotion. And this is Opus 4.8 versus Gro 4.5. I would say Opus 4.8 won by a long way here. It just looks nicer. It's smoother. It's more interesting. It's faster. It keeps your attention more and

the colors look nicer, too. So, if I had to pick between them, I'm going to go with this one. Then we have the parachute game. So, this is Gro 4.5. Opus 4.8 didn't work at all on this one, as you can see. Gro 4.5 works a little bit. I don't really like the detail. I wouldn't play this at all, but if you want to use it, you can. And you can see like it actually does countdown

the altitude and everything else, which is pretty nice. Next up, we have the flight simulator. Now, this one I actually found really difficult to play. Like, you see how it's just a bit buggy? You can't really move forward or backwards right here. Like, it doesn't make any sense. Let's have a look at the Opus 4.8 version. I would say that's a lot nicer. Yeah, a lot more interesting, cool colors, bit buggy in places, but at least

you can actually play it and do something interesting with it. So, it looks like Claude is making a comeback here. One thing that I did actually personally test as well is using Grock 4.5 inside Hermes agent. If you're using it directly in the chat, it is very very slow from what I saw. It can take like 10 seconds just to reply on a basic chat message and then it's it doesn't feel very agentic is what I'm

saying. So I probably wouldn't use it with Hermes unless you're for example using it for uh generating video or images and that sort of thing then Hermes is pretty useful with it. So we have Opus 4.8 eight with the neon arcade game over here. And then we have over here Grock 4.5 which is like a 3D fun game to use. Looks super nice. Nice UI etc. More interesting. I would say again Grock won that round. So

it's currently 7 to three. Grock is crushing it so far. Let's see what we get next. So, we got another flying game. It's pretty nice. I'm not really sure how to use this or play this properly to be honest with you. It's not very obvious to me, but actually, no, you kind of get the hang of it. It's It's all right. Not bad at all. Let's have a look at the version from Opus 4.8. See, like

there's not much going on here in terms of Opus 4.8. like there's no characters, there's nothing going on, there's nothing flying towards me. This one has a lot more interesting, entertaining elements. Then we have Dragon Flight. It's super hard to control this. I don't really know what it's supposed to be doing. It's just like flying back and forth, which is a bit confusing. If we have a look at the OPU 4.81, I would say this a

it's a lot more linear, but it actually works and it's kind of interesting. Um, more fun to play. Yeah, I would say OPU 4.81 that one. So it's now four to eight. This one for sure like Grock 4.5 just generates 3D things by default. If we compare that versus Opus 4.8 nowhere near the same, right? Nowhere near as interesting. So this is some other stuff it's created. Let's have a look what we got next. Next up

we have Outrun. So this is Grock 4.5 again. Really cool. really fun to use. This is Opus 4.8. Pretty buggy. Not that great. And then we have the pool game. The angle of this is really weird. this game if you look and it's pretty hard to play that. Let's have a look at the version from Opus 4.8. It's It's a bit more 2D, but it's easier to use, easier to control. So, I'm going to go with

Opus 4.8 for that. Then we have this maze style game. pretty cool on on Gro build. Uh, sorry, Gro 4.5 looks pretty nice. Yeah, actually works. Fun 3D game looks cool. Let's try out now Opus 4.8. It's not bad, but you can't control it in the same way. It's it the camera angle is really difficult to control as well. So, I mean, overall on the actual builds that we've tested, I would go with Grock 4.5. However,

what I will say here is just in general when I've tested them both out, like if I was for example building out a agent operating system, I would still go with Claude over 4.8. And that maybe that's just me being biased. Maybe that's my brain trained to using something like Claude, but I honestly prefer to use Claude for for building stuff like this out. So actually for building systems, for building workflows, you know, for work related

stuff, then I would go with Claude Opus 4.8 Okay, for a gentic stuff, you don't really get a choice because for example, if you're using Hermes agent, you can't plug ow with Claude anymore into Hermes agent. And then when it comes to like building out fun projects like we were talking about before, you know, the the games that we created, Grock 4.5 is very very good, like holds its own. So what I'd recommend for you is

that you test out. I probably would wouldn't use Grock 4.5 for aic stuff unless you're using the images and videos. And when it comes to Claude, Claude is still really powerful for using inside Claude desktop and building stuff. So everything that you're seeing inside the agent OS, we built with Claude and it's it's a really powerful system that we literally built from zero. So thanks so much for watching. If you want to get our full agent

OS with Grock and all the Claude systems that we've put in there, for example, like the video agent, the SEO agent, the uh Grock 4.5 Hermes Oracle system we've got inside here. You get it all inside the AR profitable room link in the comments description or just go to the arrum.com. Inside the community, you can ask questions, get help and support in real time. I personally answer all the questions inside there. Inside the calendar, you can

jump on weekly coaching calls and share your screen, meet other members. Inside the classroom, you can get access to all of our best trainings. And you can get the new daily updates with the agent OS system. You see when it was last updated, the video tutorial and a full step-by-step guide and a zip file to install it. And then we have new daily tutorials like you can see right here based on what's actually working inside the

map. You can meet people in your local area who are building with AI agents like you. You can get that all inside the AI profit boardroom link in the comments description or just go to the profit.com. Thanks for watching. All right, peeps. We've covered Grock 4.5, Hermes agent with Grock 4.5, AISO with Grock 4.5, and also the Agent OS latest community questions, plus Grock 4.5 versus Opus 4.8. Thanks for watching everyone. Appreciate all the questions and

I will see you on the next one. Cheers. Bye.