Sonic 5 Tested, Fable 5 Returns: Model Comparisons and Agent OS Updates

All right, we got Sonic 5 out today. Fable 5 export control has been lifted as well, so we're going to expect that to come out soon. And we're going to test out Sonic 5, see how it performs. We'll be testing out versus GLM 5.2 as well, building out some cool stuff in the Agent OS. And if you got any questions, feel free to ask. Good to see you. Thanks for joining. Interesting. Interesting. I think we're all

excited for it. I think that's I think Fable 5's coming back now, isn't it? Export control lifted. Sonic 5 just dropped. I think we're looking at next what 24 48 hours, something like that. Let's see. Wow, that's wild. Yes, it's on then. Can't wait for that. Heat. Heat. Start doing the goldie bench tests for Sonic 5. Start with these and do the rest. Research all the benchmarks on Sonet 5 cost etc. And then do a guide

on this and also show side by side how they perform with a few different tests. Heat. Heat. All right. So, we got a question here. Do you have any views on Sonic 5 versus Opus 4.8 for knowledge base? So, I'm testing out right now, but from what I've seen, um, in some ways, Opus 4.8 is better because I think it's cheaper and also it performs better. And even like at max, if you're using max with Sonic

5, from what I've seen, it doesn't perform as well as Opus 4.8 or it's just kind of like level. So, from everything that I've seen, the hype around Sonic 5 is not that good. I'm going to test out today personally, but from what I've seen on the benchmarks, the costs, everything else, I probably wouldn't switch at this point. However, if Fable 5 comes out, then of course light switch cuz that's that's the best. So, from what

I've seen so far, like I wouldn't switch to Sonic 5 at all yet. Just wait to see what comes out later this week. Thank you. Welcome here, Tony. Thanks for joining me. Heat. Heat. Hey hey hey. Heat. Heat. N. Heat. Heat. N. Hey. Hey. Hey. Hey hey hey. Heat. Heat. Heat. Heat. Hey hey hey hey. Heat. Heat. David says, "Which one is the best to use?" I assume you mean which AI is the best to use.

So, you don't spend much. So, if you check out Goldie Bench, you can look at the free local models and from there you can decide which one you prefer the most. Um, yeah. I mean, like, as you know, if you're looking for the best model, but you're also looking for like the free or cheap one, well, usually those two things don't go hand in hand, right? So, you've either got to choose. Do you want to pick

the best one or do you want to pick a mediocre one that's cheap? and it sounds to me like you're going for the cheap one. So, I would go with that. And then I think if you want a cheap model that's right up there, go with GLM 5.2. Cheaper than Claude. Still pretty good on performance. You can see the leaderboards on Goldie Bench. Um, yeah. And I would stick with that. Heat. Heat. Heat. Heat. Heat. Hey

Heat. Heat. Heat. Hey hey hey. back in a sec. Baby, Heat. Heat. Hey hey hey. Hey, baby. Nat. Heat. Hey, Heat. Heat. Heat. dictator. Hey, hey hey. Heat. Heat. Gemini, Super Grock, and Float. Nice. Um, yeah. I mean, Gemini I I don't really use honestly these days. Super Grock pretty good. Probably Grock build is is one of the best that came out recently. I was feeling the reggae. This techno isn't bad either. Quick question. Why would

you say to integrate Obsidian? Um well, let's say for example, you have Hermes agent builtin memory. How do you share that with Claude? How do you share that with the new CLI you set up? Right? You can't really cuz it's just for uh just for Hermes. So they're not going to run like natively. So that's why instead what you want is a system where you've got Obsidian shared across all your agents and then you're not just

thinking in terms of one agent cuz like if you saw for example last year it was all about Claude. This year it was about open claw in January or February. Now it's about Hermes agent. It's like the agents and the harnesses change all the time. So instead of relying on the agent's memory, build your own memory so that you can switch and swap and change no matter what happens next. Thank you so much. Rocker says, "Hi,

huge fan of your work, buddy." Thank you very much, sir. Appreciate that. Do you know I'm going to put the reggae back on as people are feeling that. There we go. Any requests on the music, let me know. We'll get the agent on it. Can you build a short video generator inside the notebook section so that we can just generate one in a couple of clicks? Make it super easy. Thank you. Glad you like that. So,

Empire, welcome back, sir. Hello, Julian. How is Claude Sonic 5 by now? Nowhere near Fable 5, right? You guessed it, right? We're actually we're testing it right now. So, we're testing Sonic 5 versus GLM 5.2. And from what I can see so far, GLM 5.2 two is beating Sonic 5 4 to one. Now, we'll continue all the tests. We'll run it on Goldiebench. I'll create a guide about it in a second. But right now, it looks

like GLM 5.2 on actual actual tests, right? Not like, you know, theoretical benchmarks. On actual tests, what we're seeing so far is that GLM 5.2 is beating Sonet 5. However, on aentic benchmarks, if you really care about those, Sonic 5 is winning. But let's see. But yeah, not looking good for Sonic 5 so far. However, on these tasks, he's doing okay. So, you see, uh, he's scoring about an average of seven seven out of 10 on

our tests for Sonic 5. So, it seems like it's okay, but it's not as good as uh GLM 52, which I think we all kind of knew. But from what I've seen, Fable 5 might be coming back soon. So, we'll come on to that in a sec as well. Heat. Heat. Heat. Heat. So if you want a cost effective stack for SEO, I'd recommend uh so I mean if you want to keep the cost down, right,

you can use Hermes agent. For the link building outreach, you can use Opus 4.8 for creating the content cuz it's obviously included in the plan. So, that's not too bad. Otherwise, if you're really on a budget, you could use GLM 5.2. And then for rank tracking and keyword research, you could use a combination of Google search console, which obviously free, and then also open SEO, which uses data for SEO, which is a lot cheaper than using

something like a dress. So that's the way that I would look at it. Wow. Oh, we got big announcements on Fable 5, my friends. Let's go. Daddy, daddy. Heat. Heat. Rocker says, "Hi buddy, is there an AI tool I can use to promote my website or create back links so that Chatch GPT can show my website in its searches?" Yes. So, you want to check out the agent OS. That's what we're using to grow our website

traffic. I've got loads of tutorials out there, but with the agent OS, we can look at what keywords we could go for. So, it can do the keyword research, you can create the content, deploy it to our websites, and then also you've got Hermes agent that can do the link building for you. So, you you can basically automate all three of those things using the agent OS system. Heat. Heat. N. Daddy. Yeah, normally around this time

each day. Hey hey hey. Heat. Hey, Heat. Yeah, I'm already working on the guide. I just mentioned it a few minutes ago. GLM 5.2 is pretty good. It is pretty good. I think if I had a choice at this point, I'd just go with uh GLM 5.2 over Sonic 5. Okay. Nice. Hey hey hey. Hey, Hey hey hey. Heat. Heat. Daddy, daddy. Heat. Hey. Hey. Hey. Heat. Heat. empower Julian what free API I do you recommend

B now and open router out alpha for AI agents is still valid or are there better ones? So I would test them all yourself. It depends what you're trying to do, right? So if you're trying to code something like North Mini code is actually not bad at all. You could use Gemma 4. I think I don't know if our alpha is still available for free on open rout because they actually came out and uh someone usually

with stealth models like that you don't know who's created them but there was I saw a tweet the other day that was explaining who it was. Um but yeah so I mean like Gemma 4 you've got uh there's actually a free models routter as well on open router. They're all a bit looking at the models that are available for free. Yeah, probably North Mini Code Per 3 as well. I've not tested that, but that could be

an interesting one to try. And you got Neatron 3.5 as well. I would again I would test them, see which one you like. This is a real one. Uh yeah, Anthropic have just announced that they're releasing Fable 5 back tomorrow. So, you know, within 24 hours, we should start seeing access again, which is exciting. Good AI and good rhythm. That's what we're talking about. Cheers. Heat. Heat. N. Honestly, From what I'm checking, and I'll do a

guide on it in a second. I'm just preparing that right now. But from what I'm seeing so far, I wouldn't uh I wouldn't use Sonic 5. You know, I've not when I've been coding this morning, I've not switched over. I have tested Sonic 5 relentlessly today. Um but it's not it's not as good as you think. That's for sure. I think you you know, you're looking at Fable 5 coming out within 24 hours. I would go

with that. No, that's a lot better. And I'll show you some stuff that we built with Sonic 5 and also how it compares. Plus, you should see some of the reactions on it. It's pretty wild. Uh I think a lot of people are not happy with Sonic 5, which is fair enough from what I'm seeing so far. But I'll show you what we've built, how it works side by side as well compared to GM 5.2. So,

we'll come on to that in a minute. I've literally just been testing it. I think it dropped about 5 hours ago, 6 hours ago, something like that. So, we got a question here, which is how do I link chat GPT subscription to Hermes? So, what you would do is you're going to go into your terminal. You're going to run Hermes model as the terminal command and then from there select chat GPT as the model and then

from there login with Oorth. A wolf is what you need. Happy Oh, hey. Yeah, if you want to get the Agent OS, it's inside the AR profit boardroom. So, feel free to check it out. We got a question here which is am I correct in thinking you can't link claud subscriptions simply with a off. But if you got chat GPT, you can use Chat GPT's models with Hermes. Exactly. So on Hermes, you can't use Claude Of

you can use Chat GPT. You can also get a coding plan. Uh so for example, you can get GLM 5.2 as a cheap coding plan. And then you can plug that into Hermes agent as well. And then if you're already subscribed to Twitter, you can actually use your Twitter and gro a with Hermes agent as well. And that way you kind of get two in one because you get Twitter and you also get Grock. >> Yeah,

thank you so much. I mean, we spend a lot of time on it, you know, when I'm building the agent OS, the agent OS out. I mean, you're looking at 3 to four hours a day that spent building it out. I don't know anyone else who spends as much time building this sort of thing out and improving it every day. So, yeah, I think it will help you a lot. You've probably seen all the testimonials and

wins of people who are using it. So, it's a great system. Literally changed everything for me, I would say. And it also makes AI a lot more fun to use if you have an AI, you know, if you have an agent operating system. Yeah, it's huge amount of work. If you want something that's actually really good and really smooth and polished, it's a lot of work, but it's worth it. You know, like the thing is, if

I create something like that and then let's say we've got 3,900 members inside the air ballroom and they're all using it or even if 50% use it, then it's 100% worth it. But if it's just you creating for yourself spending all that time improving it and everything else every day, it's a lot of work. So, I think it's better if like you get it from someone else who's, you know, working full time on it. Nice. This

guide needs to be a lot more visual. So, I would recommend that you pull in the traffic statistics and graphs from our latest SEO projects, the ones that are doing pretty well. I'd also recommend that you show more visual stuff inside the content and also add diagrams as well. Right now there's no diagrams, there's no examples, there's nothing. There's not even links to other blog posts. So you need to make this guide a lot more visual.

It's How you getting on with the rest of the tests? Thank you. Hey, hey hey. Hey you. Yeah, I don't use a PC, so it's not my specialty. I'll be 100% honest with you. But I've got a tutorial that just dropped yesterday on how to run free models with the Agent OS. And uh you know, whether you're using a PC or a Mac, I would recommend the Agent OS system. I think it's the best. All right,

back in a sec. Heat. Hey Heat. Happy. Heat. Heat. Heat. Heat. N. Yeah, I would imagine. So if the export controls are removed, that means everyone in the world should be able to use Fable 5 again. I don't think it'll be limited to the US. I think it's all good. So, happy days. It's looking good. It's looking uh pretty exciting. Yeah, that was from June the 12th, but Now the export control has been lifted, right? So

it's all good. So I think that was before, but now it's been fixed. Not sure exactly honestly, but honestly we we don't go over the limits for Clawude on our existing plans. So I don't think it's like you know I think we've done it in inefficient way really. So I don't think we've we've had to worry about tokens too much. Big news today as the export control on Fable 5 and Mythos 5 has been lifted. You

can see this tweet by Leo right here. Howard Lutnik has officially lifted the export control on Anthropics Fable 5. Happy days, my friends. So, you can see the announcement right here, June the 30th, 2026. And according to Anthropic, it should be restored in access tomorrow. That means we should all get access to Fable 5 again tomorrow or within the next sort of 48 hours, I would say. So you can see the official announcement from Anthropic here.

They've said, "We've received notice that the Department of Commerce has lifted export controls on Claude, Fable 5, and Mythos 5." And they've said, "We'll begin restoring access tomorrow. And we'll share an update soon." This was just posted like about an hour and 20 minutes ago. So it's very very recent news. They said, "We're grateful to our users for their patience and to everyone who worked with us on redeploying the model." So happy days. We are getting

Fable 5 back and I think that's a move in the right direction. Also seems like because the the controls have been lifted, I would expect that everyone in the world gets access to that. I don't think it's going to be limited to a certain location or just for the US. I think that'll be available for everyone around the world, which is fantastic news. And you know, Claude have just released Sonic 5. Honestly, it's a bit of

a flat release. It's not that great. So, I think that Fable 5 a huge step up because if you look at the hierarchy of things, you got uh Sonic 5 down here, then you have probably Opus 4.8 just above it and and the next version of Opus will be even better. And then Fable 5 of course is coming out as well. And I think as well, you know, when you look at it, it's such a relief

to know that we're not being gated on access anymore because I think, you know, a lot of people are worried about there's going to be there's going to be a situation in the future where we don't have access to frontier models where the people who do have access like the tiny 0.01% 0 1% run ahead because they've got access to everything and we don't and there becomes this sort of divide which you know that would have

massive uh massive consequences across the world. So it's a good move in the right direction. So let's have a look and see what we got here. So this is Howard Lutnik. He said, "Over the past 2 weeks, we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US government and strengthen America's leadership in AI." So Howard is the one who who lifted the export control on Fable 5. 41st United

States Secretary of Commerce. Happy days. Happy days. That's what I say. So just to be clear here, Fable 5 should be coming back tomorrow. Um let's see what else we got here. So Susie Wilds, chief of staff, also announced under Trump, President Trump's leadership, the United States is the is the undisputed winner in the AI race. Um, and she said, "My gratitude to companies across industries who continue to work together, blah blah blah." So that's just

kind of like a general update. I mean, make of that what you will. And uh yeah, it's exciting times. It's good to see that back. I think that's a huge relief. I'm excited to see it back. In terms of what happened, you know, it came from the White House, then commerce secretary, then reporters, then anthropic, all inside like half an hour. So like 6:26 a.m. this announcement comes out. Then the commerce secretary mentions about Fable 5

out loud. at 6:31 a.m. my time. Then a reporter announces the export control has been lifted and then Anthropic make it official. So in the space of like 20 minutes, all of this got announced which is pretty cool. Now if you want if you're wondering okay what happened with the previous timeline when it came to Fable 5. So it went dark and then it came back. So it got pulled. The US government flags security concerns with

Fable 5 and Mythos 5 and access is cut and that was on the 16th of June. And then we were all going back and forth. We had an Axius and Forbes report that Return is on track for Fable 5 and could come back this week. But the weekend passes with no signature. That was the 27th to the 27th uh 29th of June. 1st of July, White House announces it, then commerce approves, then we get Anthropic saying

access is back uh within 24 hours. So happy days there. Nothing on GPT 5.6 by the way. Not heard anything about GPT 5.6 coming out or Soul or Luna or anything like that. So that seems to be stuck in preview. I mean if you asked me right is GT 5.6 going to be better than Fable 5? I would say no. I think you know Claude have consistently over the last 12 to 16 months outperformed GPT 5.6

And I think that'll continue. Uh sorry, they've outperformed Chat GPT. I think that will continue. Now, the thing that I would say is whatever happens next, focus on the system. Don't worry about the model because, you know, you've got Sonic 5 coming out today. You've got Fable 5 coming back out tomorrow. You'll have something else next month. The way that I'm looking at this is we'll get Fable 5 back into our agent operating system and we'll

probably make that the default CLI, but whatever happens next, focus on the system, right? And that's why we build out agent operating systems like this with a shared memory and everything else because these models swap and change and some will be in preview. It won't be the first time that a model gets pulled. In fact, I think that'll be a very consistent theme over the next 12 months. And also, I don't think it'll be the first

time that a Frontier model goes into preview and nobody can access it. So, you know, just be prepared for that in the future as well. And and that's why we build great systems instead of worrying about the models and chasing each model cuz for most people as well, you don't need the Frontier models. You just need a good system to automate most of what you do. So, let's see what happens next. Fable 5 is coming back.

It's official. It was based on a policy, but it is a warning that you know any model can vanish overnight. So don't put all your workflows into Fable 5. Again, be balanced, have great systems, diversify. And the main thing I would say here is the system wins, right? One dashboard, every model, and then you can switch everything around. So if you want to get our agent operating system, you know, Fable 5 is one model. The Asian

OS is the operating system that runs any of them. Fable 5 local free all from one dashboard so that an outage like this never stops you. You get one dashboard for every model. You get every CLI plugged in right here. You get Hermes open claw clawed all plugged into one system with a shared memory. And every time a new update comes out for example like today notebookm just dropped a short video. We already plugged that into

the agent OS so that you can get the most out of these models as well. So exciting times for Fable 5. Hope to see inside the AR profitable boardroom. We actually got a lot of training on Fable 5 inside the AR profitable boardroom too. If you search the classroom, you can see loads of training on Fable 5 and also how it performs versus other models and systems like for example Fusion and Mixture of agents. Now, this

is my community focused on helping you save time and scale with AI automation. You can post questions inside the community. I answer them personally each day. You can jump inside the classroom. Get access to all of our best trainings. And you can also get the agent OS which we update daily with vid a video tutorial, a step-by-step guide and the zip file to install it. Plus new daily guides. It's useful stuff actually comes out. Plus inside

the calendar you get four week coaching calls. You can meet people in your local area from the map. And that's all available inside the AR profit boardroom link in the comments description or go to the arprof.com. Uh, sorry about that. Yeah. Uh, share the right screen next time. Noob mistake there. Noob mistake. Let's make sure we get that fixed. Still though, either way, happy days. You'll come out later today, sir. Yeah. Thank you, sir. Appreciate it.

So we got a question here. Which model do you recommend using for the voice activated Hermes agent inside the agent OS? I'd recommend chat GPT real time. It is by far the fastest and most responsive voice API I've seen out there. And I I actually only found that out because a friend was creating a voice agent for their clients and they showed me and I was like, "Wow, it's it's literally instant." So, I'd recommend that. We

did try 11 labs, nowhere near as good. Did try Miniax, nowhere near as good. But chat GBT realtime, amazing for the API. So today we have the release of Claude Sonnet 5. Apparently the most agentic model from Claude Ananthropic yet. And apparently you can see the announcement here. It just dropped a few hours ago. It can make plans uh use tools like browsers terminals run autonomously at a level just a few months ago required larger and

more expensive models. We'll come on to this in a minute. I'm not going to hype it up here because uh you'll see from my test. I'm just going to tell you the honest truth. So here you can see you got Sonic 5, you got Sonic 4.6. So it is a step up from uh Sonic 4.6. If you actually look like Sonic 5, nowhere near the same level as Opus 4.8. So if we look at Agentic Coding,

63% versus 69% for Opus 4.8. I mean literally on every single benchmark here, Opus 4.8 8 is crushing it. Now, if you were comparing 4.6 from Sonet versus 5, then of course it's a big step up, but I think for most people, like you you're probably still going to stick to Opus 4.8 for building stuff, especially at this point. Now, we can expect Fable 5 to drop very soon. Anthropic just announced that they're going to be

releasing and giving people access back again within 24 hours for Fable 5. So, this kind of just feels like weird release where I don't know if many people care about it. I mean, if you're watching this, do you use Sonic 4.6 already? Are you not just sticking to Opus 5 uh to to Opus 4.8? Are you not going to just switch to Fable 5? For me personally, I just don't know if this is really a big

update or not. Now, if we actually have a look at what we've created so far on Goldie Bench with Claude Sonet, let's take a look right here. So, if we pull up the benchmarks over here, it did create some pretty cool stuff. And I'll show you what we've built, how it works, etc., how it performs. So, we can gather some examples side by side over here. Here's a couple of examples. So, this was like a a

raycaster maze. I mean, it can create pretty nice stuff, right? Like, this is pretty cool. The graphics pretty nice. The colors pretty nice. I mean, it works perfectly as a game. For some reason, the controls were a bit backwards when I test out personally, but apart from that, we're all good here and it's working nice. Now, if we have a look at this one, this is kind of like a little test. The orbit test for a

galaxy didn't work at all. So, some things have broken on it completely. And then this was kind of like a a cool synth wave background as well. Then, we actually created a crypt game as you can see right here. And it looks pretty nice. play. I mean, it's again, it's smooth, easy to use, etc. Not bad at all. But when we actually look at it on the benchmarks, you'll see how it compares in a second. You

also might be wondering, okay, how does it compare versus G 5.2? I'll do a deeper tutorial on this later, but basically, if we have a look side by side, this is the maze example from Sonet 5. Pretty nice. I do think it's a big step up versus what GLM 5.2 created over here which is you know using the same prompt but it's just a little bit buggy not quite as smooth um you can see I mean

a little bit buggy I would say very buggy at this point so when we've compared it on tests like that it didn't do so well but then there were other things where for example the orbit didn't work at all with Sonic 5 but if we have a look at GM 5.2's version it actually works really nicely and it can complete the task. So, there were some tasks where Sonet 5 actually totally failed and then GM 5.2

crushed it. But we'll come on to more stuff like that later in a different tutorial. Now, we can also see here, for example, side by side, we've got two outputs from Sonic 5 side by side for this game. As you can see, and also you can see it didn't do so well in terms of public reception. So for example, you can see this uh tweet from Lean who said sonet 5 goes straight into the garbage bin

1 1.2x more expensive than Opus 4.8 8 Max, two times more expensive than GPT 5.5x high, five times more expensive than GM 5.2. I mean, it's pretty wild when you put it like that. I can understand why a lot of people are quite negative about it cuz basically Opus 4.8 is outperforming Claude Sonic 5 on many benchmarks as you saw earlier, but Sonet 5 is more expensive than Opus 4.8. So, Opus 4.8 is better, but Sonic

5 is more expensive. How does that make sense? And again, I think with Fable 5 coming out soon, this is going to be one of those things where people don't even talk about Sonic 5 anymore cuz it's just it's been received so badly. You can see another tweet here from Bridgemind. So, he said, "Anthropic seriously messed up with Claude Sonic 5. The token efficiency is so bad, it actually ends up being more expensive than Opus 4.8.

What are they doing? Uh, so it's been quite a brutal day for Anthropic. Here's another one from Bridgem saying, you know, Claude Sonic 5 looks terrible on Cursor Bench. The whole point of using Sonic is so it's faster and cheaper. Um, but if we actually look at the leaderboards here, we can see that Sonic 5 is way underperforming Opus 4.8. So if you're going to ask me like, okay, what is the best API to stick to?

What model would you use with Claude Code or Claude Desktop, whatever? Just stick with VPUS 4.8. Like, there's no point using Sonic 5 from what I can see right here. Especially when He's scoring so low on Goldieb. I mean, let's compare it to a couple other things. So, we've got Claude Sonet Fugu Ultra. So, Fugu Ultra is the new model from Japan. So this is the output for the crypt game from Figo Ultra. And basically what

it does, it fuses two models together. And you can see like the outputs are pretty nice from that. Right now, if we have a look at the outputs from Sonet 5, you can see like the colors a lot darker. There's not as much going on. It's not as interesting, etc. So even like the models where you fuse older Frontier models together are way outperforming uh Sonify by quite a long way. Like this is just a much

more interesting uh response. Let's have a look at another example. So we've got liquid in a bowl here, right? So what should be happening here is that you have like liquid in a bowl and here's an example of what it looks like from Sakana Fugu. Now, if we compare that to Sonet 5, just doesn't look as good, right? And and that's a huge gap in quality. Again, if we compare the the orbit outputs, Sonic 5 totally

failed. Sakana Fugu crashed it. Now, we're going to do more tests. And it just dropped out a few hours ago, but from what I can see, I just I wouldn't use Sonify. I don't see the use case. It might be great at writing potentially. That might be something. But apart from that, yeah, there's no need to to switch or anything like that. And you're probably asking, why would you release a worse model that's more expensive? I

don't know if like for example Anthropic just felt the need to to like bring out something new because there hadn't been a new version of Sonic for a while. I'm not sure what they were doing here or whether they were just disappointed themselves with the outputs. But from everything I can see, I just, you know, I wouldn't I wouldn't use Sonic 5. What I do like what I do actually appreciate is that they were so honest

on the benchmarks cuz there's probably a lot of companies who would try and make Sonic 5 look better than it is. But from what I can see, like the the benchmarks were ruthlessly honest. The Opus 4.8 is a lot better. Either way, whatever happens, you know, when there's a model launched like this, focus on the system. Don't worry about the models. Fable 5 will come out pretty soon as well. And so, you know, I wouldn't worry

about like what's the hot model of the week. You want to have a system where you can use something like the agent operating system and it doesn't matter what comes out next. So, for example, for us, if Sonic was 5 was actually good, we would plug it into our agent operating system. It's not. So, we'll keep Opus 4.8 instead. And that's the whole point here is like whatever happens, you want to have systems ready to go

that are flexible no matter what comes out next. I mean, for example, something else that we did, we plugged in GLM 5.2 into Claw Code so that we could get the best agent harness with one of the cheapest models and then we can combine the best of both. We've got like a workspace where we can track everything. We've got a memory system plugged into all of this. We've got Hermes agent ready to go and claw CLI

and everything else. And then we've got a team of agents inside our group mastermind and everything else. So the point here is like focus on the system and having a great systems and then if good models come out you can swap them in and if they don't like this you're not disappointed because you still got great systems and it doesn't matter what comes out next. So if you want to get our agent operating system is inside

the air profit boardroom. This is our community that helps you save time and grow and scale with AI automation. Inside the community, you can ask questions, get help and support whenever you want. Inside the classroom, you get access to all of our best trainings. If you want to get our agent OS system is over here with a video tutorial, the step-by-step guide, and the zip file to download and install. And then you can see there are

new daily tutorials right here. Also, if you want to jump on weekly coaching calls, we've got four coaching calls per week where you can share your screen, ask questions, get help and support in real time. And inside the map, you can actually meet people in your local area who are building with claude and agent operating systems and stuff like that. But overall, Sonic 5 not great wouldn't switch. Continue using Opus 4.8 and let's hope Fable 5

comes out soon. Thanks for watching. Let's see what questions we got here. DG, great. Thanks. I have quite the latency, so I'll use GPT. Hey, bro. What do you recommend doing to reduce usage in claude? I would recommend using headroom. Headroom is pretty good. It's free and open source. You can plug it into whatever CLI you prefer. Where is open claw? I I think just people don't use openclaw is I I mean I basically created tutorials

on both of them and covered like the full journey between when openclaw came out and then also when Hermes got released. And then I remember testing out Hermes for the first time and thinking actually this is pretty cool. And then gradually Hermes just got better and better. More and more people started switching to it until eventually nobody really uses open claw anymore, right? Like I think that Hermes is just so powerful. I don't see anyone using

Open Claw. And also bear in mind like Open Claw it was just breaking all the time. Like every single update that came out, Open Claw tended to get worse. And maybe they fixed it now, but I think we're at the point now where it's just like, okay, why not just stick to to Hermes as well? So if you look at open router like the trajectory of Hermes is going up like that and the trajectory of openclaw

is just dropping and that's because overall Hermes is a much better model much smoother model. Exactly. That's a really good point, Empower. Fantastic point there. That that sums it up perfectly. Mac says, "I used it and it seems annoyingly slow." Would agree with you. Would agree with you. Empower says, "Thanks a lot. You're the best." Thank you very much, sir. And DG says, "Also, when using Sonic 5, it can switch to Opus depending on the prompt,

so the token cost can fluctuate according to that propping." Yeah, that's kind of annoying as well. I actually saw that with Fable 5. Like sometimes you just randomly switch and you're like, "Why are you doing that, mate? What are we doing here? All right, we're going to get more of these tests done with Sonic 5. Can you update the website as soon as the new tests go live? Heat. Heat. Hey, hey hey. You're daddy. Hey hey

hey. Hey, hey hey. Heat. Hey. Hey. Get out. Creative Cheeto says, "Does Hermes use the official open AAI or does it use chat cheap potato off?" You can use chat cheap off. When you go to Hermes model inside terminal, select codeex or opening eye and then select o. And from there you can use oorth Claude Sonic 5 versus GLM 5.2 the oneshot showdown. Who wins? We're going to walk through it today and side by side. We'll

be comparing how Sonic 5 compares to GLM 5.2. So, let's get straight into this. And the first thing that we're going to start with is a crypt game like a dungeon crawler that we've created with both of these models. So this is GM 5.2. This is Sonet 5. Which one wins? Let's compare them side by side. We'll also compare the benchmarks in a second as well. So if we have a look, this is the output from

GLM 5.2. And uh not bad. Not bad at all. Let's have a look at the output from Sonet 5. I'd actually say the graphics are smoother, but the gameplay itself is not that great. Like there's literally nothing going on here. It's super dark and there is no game to be played. It's just you walking around a maze. So, not that great. And I would say this is kind of the theme that I've seen across everything. And

we'll come on to some more tests in a second in terms of how they perform. So, also got another one here which is like, you know, build a raycaster maze. So, let's see how they perform. This is the output from GLM 5.2. And if we have a look here, you can see that it is super buggy, right? It's really really buggy as you can see here. Um whereas for example, if we have a look at the

output from Claude Sonic 5 actually does win in this particular contest. So it is possible to get better outputs with Sonic 5 than GLM 5.2. Bear in mind that Anthropics Sonet 5 is closed whereas GLM 5.2 2 is open source, so that plays a big difference as well. And also, of course, GLM 5.2 is a lot cheaper than Sonnet 5. So, let's see how they perform on the test here. If we compare them on Cursor Bench,

well, Fable 5 max right at the top, extra high. Uh, Fable 5 is absolutely dominating, crushing here. If we have a look at Sonic 5, it's way down the charts. Uh, you can see here, for example, it's number 13. However, GLM 5.2 is actually a lot lower. So, you can see it scores 61.2% for Sonic 5 on Cursor Bench, whereas Sonic 5, sorry, GLM 5.2 scores 54.6. So, there is a big difference here in terms of

how they perform side by side. Now, you also might be wondering, okay, should you pick Opus 4.8 or should you pick Sonic 5 if you're using Claude Claw Code, etc. So, what we can see so far is that Opus 4.8 a crushes sonet 5 on the benchmarks as well. So if you have a choice between them, I had to go with Opus 4.8. But bear in mind that Fable 5 is expected to drop within the next

24 hours. That's been announced officially from Anthropic. So, you know, if you had a choice, I'd put Sonic 5 down here. Then we've got Opus 4.8 above that. And then we have Fable 5 that's even better than all of them. And also something to bear in mind as well, you see in this tweet by Lisa, one point it's 1.2 two times more expensive than Opus 4.8 and Sonet 5 is two times more expensive than GF 5.5,

but actually five times more expensive than GLM 5.2. So bear in mind like if you're using the coding plans, so you can get a coding plan with Claude or you can get a coding plan with GLM 5.2. If you're using the coding plans, it's going to be a lot cheaper with GLM 5.2. And also bear in mind like if you want like the full power of GLM 5.2, to but you prefer using something like clawed code.

We actually set up something called GLM code inside the agent OS system so that we can use the power of GLM 5.2's brain inside claude code and then we can say everything that we've created right here and it performs pretty well when we're doing that as well. So you get like the agent harness of clawed code but you can plug in GM 5.2 and then you know either way whatever drops whatever comes out next you can

get the most of these models by having great systems like this. Now let's have a look how it scores on Goldie Bench. So obviously Claude Sonic 5 just dropped a few hours ago. It's a million token context window. It's scoring pretty well on SWB verified to be fair. They've actually said it's their most agentic set yet. But I mean that should be the case anyway, right? I would expect that anyway. Like for example, if you're using

uh the older version of Sonic 5, are they going to be as good as the new Sonic 5? No. But at the same time, how does that compare versus Opus 4.8? Well, Opus 4.8 is crushing it. That's that's the difference right here. If you actually look on the benchmarks as well, you can see that Opus 4.8 is pretty much outperforming on everything. It's just that Sonic 5 is a big step up from Sonic 4.6. That's a

big difference here. So, if you want to see some other stuff that we built with Sonic 5, let's have a look here. So, the outputs were kind of disappointing. You can see with this like diagram, it's just a little bit boring. The orbit didn't work. The synth wave is okay, but it's not doing anything special. The one thing I would say as well is like I'll keep GM 5.2 in the agent OS but I won't be

using Sonic 5. And also bear in mind like one big difference between both of these models is that with GM 5.2 Two, you can actually use oorth with Hermes agent or openclaw, which means you can use them as the brain to power your agents. So you can use them inside cloud code. You can use gm 5.2 with their CLI. You can use it agentically inside Hermes or for example openclaw. But you can't do that with claude

because they block a login which means that you have to pay for the API. Now, we did do some other things as well. So, this was uh a game that we created called Dragon Realm from Sonic 5, which is it's not bad. I mean, it's a little bit basic in terms of the responses and everything else, but it's not bad at all. Now if we compare that versus GLM 5.2 this is GLM 5.2's output. I would

say it's just slightly better, slightly more interesting. Bear in mind like you've got more characters that you're going to come across inside this game. Whereas, if you look at the output from Sonic 5, there's not much really going on, right? When we look around, there's no characters or anything like that. It's just kind of like a basic landscape. the same for example we created like this game as you can see which is quite fun easy to

play with GM 5.2 too. But then if we have a look at the output from Sonic 5 is you look at this and you think what is this? I think we created something like this 12 months ago with uh with Manis or Gen Spark. So you know that's the the common theme that I'm seeing right here. But then you look at something like this. This is created with Sonet 5 and it looks a lot more interesting

and it's a lot better than if we for example compare this versus GM 5.2's output which is just super boring on this. So it is a mixed bag like you know Sonic 5 does hold its own against this but again like I think that Opus 4.8 would do even better both of them. This was a game called Neon Drift. Kind of like a 3D racer game. This is created with GM 5.2. And then this is the

one created with Sonic 5, which is still pretty cool. I mean, it's like a 3D game, quite fun to play, etc. So, here's another one. This is a game kind of like an RPG open world. So, we called it Dusk Wanderer. And if we look at the landscape and the colors from GM 5.2, it's a lot nicer. It feels a lot better than, for example, if we compare that to Sonic 5 here. Look how basic this

is. Nowhere near the same standard or level, right? And if you want to check these out for yourself, you can check them out side by side on Goldie Bench. Here's another one. So, this was a website that we created, it feels super basic with Sonic 5, like for web design and UI. Whereas, if we compare that versus the output from Chilman 2, it just feels a lot cleaner. It's a lot nicer. I mean, it does use

the generic AI colors, the purple gradient, but everything else apart from that is pretty good. And then if we have a look here, we've also got webOS. So, this is literally an operating system. This is the version that we got from GM 5.2. And if we create something here, it seems to work perfectly. It saves this application. Then we close it. And if we open it back up again, it should have our saved notes inside notes

here. Let's have a look. Yeah. So, it saved everything that we created. And then, for example, we can open up the calculator app. And it's ready to go, right? And so, that works pretty nicely. This is GLM OS. Now, let's check the same thing with Claude. You see how the background just feels a little bit little bit basic? It's not as smooth versus this one. And then let's check the notes app. See if it actually saves.

So we'll type some notes. Close that. Open it back up. It does save it to be fair, but it just doesn't feel quite as nice. There's something about it that's missing in terms of the final finishing touches. So overall, if I had a choice between them, I'm going to go I'm going to I would stick genuinely with GM 5.2. It's cheaper. We can use it aentically with Hermes. It's open source. we get a lot of tokens

on the plan and it consistently outperformed Sonic 5 on the benchmarks. Now, if you look at the actual leaderboards here on Goldie Bench, you can see Fusion and Hermes mix of agents are still right at the top. Grock Minax doing pretty well as well. And then you've got GLM 5.2 down here with closely followed by Opus 4.8 and then Sonic 5. Right now, if I had a choice between all of these, I'm probably going to go

with Fable 5. And we'll test out again within 24 hours once it comes back. And it should be out back out tomorrow cuz the Xbox controls just got re-released um got dropped for that. So, that's going to be pretty interesting. And then, you know, whatever happens, we're going to plug it into our system. So, we're going to make the most of it. If you want to get like, for example, GLM plugged into Claude code as a

setup, we've got it over here. If you want to use GLM alongside Claude, we've got that set up inside the agent operating system over here. We have a memory system that links them all together and that's automatically organized by our AI agents. And then the the thing that I would say here is don't chase a model, chase a system. Right? So whatever happens with GLM 5.2, Fable 5, Sonic 5, doesn't really matter because we can make

systems like this that are super anti-fragile and just are very flexible to them. So if you want to get my full setup that's inside the agent OS uh in the AI profit volume link in the comments description or go to the profit.com and inside the community you can ask questions. I personally answer all the questions inside the community daily inside the classroom. You get access to all of our best trainings. We have a complete beginner to

expert course over here. And then if you want to get the agent OS we have that inside this section so you can get the agent OS is updated daily. You get a video tutorial a full guide and the zip file to install it. And then we add new daily tutorials based on what's just been released. As you can see now, also inside the calendar, you can jump on weekly coaching calls, get help and support in real

time. You can also direct message me and ask me any questions if you want to. And then inside the map, you can connect with people locally near you who are also building with agent operating systems and everything like that. So feel free to get it. Link in the comments description or just go to the aiprbot.com. Thanks for watching. Which model do you use for your main orchestrator? I think right now recently I've been using Grock and

Miniax seems to be the best. Let's see what happens. M3. Yeah, Minax M3 is pretty good. Yeah, Grock and Minia Max3 I'd say are the best right now for Hermes. GM 5.2 not a bad option too. back in a sec. Heat. Heat. N. Daddy, you Hey, hey hey. Heat. Heat. Heat. Heat. Nah. Hey. Hey. Hey. Hey. Hey. Pity. Yeah, I already done a a video on that, mate. 100%. I've never run out of tokens on this.

coming out in 24 hours, right? It's coming back in 24 hours. But yeah, happy days. Happy days, my friend. We have a brand new update from Notebook LM for short video overview. So, this is basically a way that you can generate short videos with Notebook LM. We've already plugged it into our agent operating system, and I'll show you exactly how this works in a second. Bear in mind, it's rolling out right now to Google AI Ultra

and Pro subscribers, but just to be 100% clear here, it's going to be coming to free users soon. So, you can see it'll be coming to free users soon. This is always the way that Notebook rolls out. They roll out new features to the premium members first and then it comes out to other members second. Now, if you're wondering what does it look like, what's the quality like, etc. Let me show you a couple of examples

right here. So, basically what these are is 60se secondond vertical videos, right? So, they're designed for, you know, like social media, for example, like quick short videos. And the good thing about this is previously you only really had a choice of using longer videos, which would take longer to generate. And also, like if you wanted these style videos where they're portrait and vertical, you didn't previously have that. So this is what's new with this new update.

So if we have a look at this video for example, we'll play them both side by side. To be fair, these are a couple of examples we've made and you can see here that the whole video was put together with notebookm. Let me show you some more inside our agent operating system. So what we actually did here is we have plugged in notebook LM as an MCP into our agent operating system because I think that's like

a much nicer way to use it. Now if we want to see the new short videos that we've created once we select a relevant notebook. So bear in mind like each video is related to an individual notebook custom trained on a specific topic. Then we can play the video itself as you can see right here. And it's not bad at all. It does have sound over the top as well. bear in mind. So it has a

voice over and then what it's doing is basically it's creating images and then putting them all together inside a video. So it's fully AI generated, fully hands off as well. And then the video itself is created with AI. And bear in mind this is coming to free users as well. So you can now generate vertical AI videos using notebookm for free and it's all based on the automated research that notebookm does as well. So, it's a

pretty powerful and useful system as you can see and also has um all of this put together. So, it's kind of like you've got a video team put in one because the script, the research, the generation of the video, the editing, the voiceover, it's all plugged together inside one system. Now, what I will say is when I look at this video, I'm like, well, is it as good as a human video? No, of course not. Is

it as interesting as a human video or like a video you create yourself? Not really. I mean, for example, if we put this side by side. So, this is the Notebook LM setup that we've got. We've actually plugged in something called Open Montage, which is an amazing amazing way of creating like film style videos. And if we compare that short style video here versus the outputs from open montage, these are way nicer, right? Like these look

absolutely especially this one and this one. These two look amazing, right? So if I had a choice between using notebookm or using open montage for generating these videos, I'd probably go with open montage. Honestly, I think like it's it's a better style. But bear in mind these are educational videos. So they're more based on the research that you plug into your notebook. So for example, we have all these different notebooks like one on you know let's

say for example meta uh by malt book. We got AISO agent OS over here mastery notebook etc. These are all custom trained pieces of research that we've created with no book allet. And so when we generate a video that's based on the research. Whereas for example, if you create a film like this, it's more cinematic. It's more interesting to watch, but it's not going to be that educational. So bear in mind as well, if you're creating

educational content, then you'd go with notebook. If you're just interested in films, uh, or something that's a bit more visual and beautiful, but not that educational, then you'd go with something like Open Montage. That's the difference between them. Also got something else that I think maybe might might be more interesting or better. So, we actually have a video agent and this is like an educational video as you can see. And basically the B-roll, the AI avatar,

the voice over, the editing, it's all put together in one single prompt. And I think honestly if I had a choice between this system and notebook for educational videos, I would go with this system. But if you want to use and generate in the future uh free short videos, then you can use notebook as well. The other benefit of using notebook length of course is that you can do research as well. So you can do deep

dives into the notebooks you use. You can create notebooks and new research about any topic that you choose. Everything is linked with an MCP. So you can plug it into your agent OS as well. And then also you can generate at literally like one click of a button audio overviews, videos, slide decks, mind maps, infographics, flashcards, quizzes, data tables and reports using notebook with just like one click as well. And I think that is pretty powerful.

So the videos themselves wouldn't say they're that great, but when you combine it with the whole system and what you can do for free with no bookm it's pretty unbelievable. And then what we actually have here is a system where we can plug in all of the stuff we've created with no bookm. So for example, if you look at this, we've got like this beautiful infographic that we created with no bookm. We've got like really detailed

research reports. We have for example podcasts and video uh audio overviews and then we have like the videos themselves. Now actually one thing to note here as well is that if you look at the videos directly from notebookm so these are the long form videos that you can generate with notebookm I would say these kind of feel more like a slideshow presentation right whereas for example if you look at the outputs from the shorts they're more

kind of like a bit more animated a bit more interesting so it depends what you prefer to use. I also think these look a lot cleaner though when you look at the video themselves. It's just a bit boring to watch and I don't see like many videos like this doing well on uh on social media honestly. Whereas I think this style might do better but at the same time I would still go with the video agent

or open montage for creating like you know really interesting pieces of content. So it depends what you want. And the cool thing about this as well, another use case of this is that if you're creating for example short videos or research with notebookm is like it's just great for learning. Like it is an amazing learning tool. If I was starting to learn a language from scratch, this is the system I would use. Or if I was

trying to take a complicated topic and break it down step by step, I would use notebook. It's pretty good for that. So overall pretty powerful system especially when you plug it into something like the agent operating system where we can also for example take the outputs from notebookm and then plug it into our memory and then we give our memory more context as well. The other thing I would say here is obviously it's a lot faster

than like editing shorts yourself or editing short videos. So, you know, if you were looking at creating a short video like two or three years ago, well, you'd have to script record and edit everything by hand. Might take you like half a day of video. You have to do a lot of research and your reach basically stalls because you can't keep up with everything. That would be the old way. But the new way with something like

Notebookm is that you drop in the sources. You get a 60-second vertical video with AI back. You can turn every piece of research you have into a short video and then you can plug them into your agent operating system so that Hermes or whatever can create videos whenever you want for free using this system. Now some people are going to watch this and they might be like well is it free or is it not free? So

it's coming to free users in the future. You might also say okay why would you want to generate short videos? Well it's just good for learning and also good for social media and that sort of thing. Additionally you might say which one is the best system to use? Is it nobook or is it open uh open montage or is it the video agent for generating videos? Personally for me I would go with the video agent first

then open montage and then notebook for generating videos themselves. But if I had to pick between all these systems for research then I would go with notebook. So that's what I' I'd recommend for you too. So, if you want to start using this whole system as basically like a content factory, everything that I've shown you today, the notebook, LM automations, the video agent, open montage, all of the playbooks, we've plugged into the agent operating system. You

also get a step-by-step setup guide. You get the exact notebook elimination OS wiring from what I've shown you today. You get four coaching calls a week and loads of other powerful stuff inside the AI profit boardroom. So if you go to the AI profit boardroom link in the comments description or go to the profit.com this is my community for helping you learn, grow and scale with AI automation. You might also say okay isn't the agent OS

system like pretty difficult to to use or a little bit technical but actually you can see for example so many people are getting results with this and I'm nontechnical. They're nontechnical. We can all get awesome results with an agent operating system like this. So you don't need to be technical to set up and there's so many amazing people in there that you can connect with as well. Also inside the classroom you get all of my best

trainings on notebook. So if you actually type in notebook inside the search here you can get access to loads of different trainings and how to combine it with loads of different tools and get the most out of it. So you got loads of video tutorials and trainings on that. You can also post questions inside the community and get help and support in real time. Inside the calendar you can jump on four weekly coaching calls and get

help and support on a call. share a screen, ask questions, meet the community, and then inside the map, you can meet people in your local area. So, feel free to get that link in the comments description or just go to the aiprofitbomb.com. Thanks for watching. Fable is back. He's coming back, peeps. Yeah, everyone's excited. Will you be using mixture of agents using Fable when it's back? That would be a great one. Yeah, I might test it

out. That's a great idea. Hermes mixture of agents plus Fable 5. Wow, that that's probably going to go straight to the top of the leaderboards to be fair. Uh yeah, got to try that. Heat. Heat. Heat. Heat. Daddy, daddy. Today we're going to be answering the latest questions on the agent operating system, which is a powerful system that combines all of your AI agents together into one amazing place where you can orchestrate your agents and do

anything you want with AI. So, what we're going to do today is we're going to run through all of the latest questions because I know if everyone else has these questions, you probably do too. If you're wondering, okay, what have we updated recently inside the system? So, we've got a lead generation tool with Hermes agent. We've been testing out mixture of agents, which is pretty wild as a tool actually. I would definitely recommend checking out. And

then also, we have our memory system plugged into side here, which basically automatically updates this stuff. In terms of new stuff, we've actually created GLM code. So this is a new section inside agent OS where you can have your GLM 5.2 coding plan which is pretty much up there with Opus 4.8 and plug that into an agent harness like Claude code which is super powerful. We also have notebookm and a new section for short videos so

you can generate short videos in like one single click using notebookm and that's super powerful. You can see some of the stuff that we can do right here. And also, you can pull in all of your notebookm um videos and research and everything else inside this section too. And then additionally, we've been building out more stuff into the SEO content pipeline. So, you can actually use open SEO with this now too. And additionally, Claude Fable 5

should be back tomorrow. So, when that actually comes out, we'll be adding Fable 5 as a default CLI inside the agent OS. And by the way, you can customize or change anything that you want with this system, but this is just what we're using right now. So, let's get straight into some of the questions, some of the latest questions we've had from the AI profit boredom community, what they're asking and how we can help. So, we

got a question from Michael 8 minutes ago. So, this is pretty recent one. And he says, "Xit guy looking for a direction to go with AI." Right. And he's running cursor. He's running anti-gravity. He's got a couple of vibe coding projects, but he says, "I am working on these, but I'm all over the place. So, looking to find like one focus, looking for one setup, etc." So when it comes to overwhelm or when it comes to

being all over the place, we actually have a system to help you avoid that. That's actually the first thing we have inside the start here section of the air profit boarding just because it is so common, right? I've seen so many people struggling with staying focused. And so we have a full uh guide and setup guide on how to never worry about overwhelm or being all over the place and make sure that you focus. Um it's

called our focus protocol. basically cuts out all the noise so you can focus. Now, on top of that, if you're wondering, okay, what should you focus on in the meantime? We've got two different pathways for that. So, we've got two different pathways to help you focus. Uh, but the main thing I would say is just focus on one thing at a time. Like with AI, you can do so much, but you don't need to do that

much, right? You just need to focus on one thing at a time and then go from there. So, for example, right now for me, my main focus is the agent operating system because I know if I just focus on that, we're going to build something amazing and that's just going to keep improving every single day. And I do the same for you. It's like just focus on one thing at a time. Don't let yourself get distracted.

Don't try and do like five different projects at once. Just do one thing at a time and stay laser focused. So we have a question here from Scott and he is running Claude desktop and Hermes desktop as well and he's also going to download Miniaax desktop. Now, at this point already, I would say you probably need an Asian operating system because if you're using three different tools, Miniax, Claude, and Hermes, well, that can get quite messy.

And so, you're clicking between tabs. It's quite hard to link everything together. I can see that you're already at the point where you're having trouble tracking multiple tasks. And so, Scott, if I was you, I'd definitely recommend using the agent OS instead because if you've got like three different apps with three different tools, it's going to be very messy to track everything that you do. You're going to lose projects, you're going to get distracted, and it'll

be hard to orchestrate all of it. So, when you're using the agent operating system, the great thing is number one, for your session history, you can store all of that inside the memory obsidian system so that everything gets automatically locked. So like every time we do something inside our system, it gets logged to our memory and then that's automatically organized inside obsidian and then for example if you generate something with claude well that goes inside your

workspace. The same with for example Hermes or anything else that you do and so for me personally I'd recommend that you stick or focus on the agent OS instead because it will help you manage everything. It will keep everything smoother and in one place. And if you want to get the agent OS system, it's right here. So we have a zip file where you can download and install it. Sanjito was asking about like how do you

set up agent OS on docker? Now for me personally I actually don't use docker and I've never used it to set up an agent operating system. So I just run it locally as you can see and that works really well for me. We do have members who have run it on a VPS, but I've never set it up on Docker. However, if I was trying to do it, I would follow the same advice that Jay Parsons

recommends that where you basically you can use the zip file to install the agent operating system and you can ask Claude to to put it into Docker for you and orchestrate that all for you in the back end. Andrea is asking about using GitHub to link together and create a system of 12 main AI agents plus Honcho Hermes and paper clips. It sounds complicated already and says she wasn't sure about permissions or everything like that and

also data and then but also needs to get this up and running. So for me personally, I think if you're worried about data and data protection, I wouldn't use GitHub to store a project cuz it's open source. And you know, if you saw the situation with Anthropic, it was pretty bad because what happened is they accidentally leaked their code on GitHub. So if they make those sort of mistakes, you can imagine that one of us could

easily make those mistakes too. So when it comes to linking everything together like a team of 12 agents and then for example, Paperclip, Hermes, etc., Instead of plugging that into GitHub, I would recommend that you just use like something like the Agent OS, right? And if you have multiple different teams or multiple different setups, for example, like Paperclip, well, that just slots in easily as you can see inside the Agent OS instead. And I think that's

a much cleaner, smooth way to link everything together and to have in one place. Also, we've got some great news today as Anthropic have announced that Fable 5 should be restored with access to us by tomorrow. So, I think within 24 hours, we'll be looking at Fable 5 coming back, which is great. And again, I will be plugging that into Claude as soon as it drops. There was an interesting post from Allan about Sonic five versus

Opus 4.8. For me personally, I'm going to stick with Opus 4.8 and Fable 5 is obviously coming as well. So, I kind of feel like this Sonic 5 is almost one of the worst releases that we've seen with AI, but at the same time, Fable 5 is amazing. So, I don't I have nothing against anthropic. I think they're fantastic, but I do think that there's an issue with Sonic 5 for sure, and it's just there's no

point using it. We also compared GM 5.2 versus Sonnet 4.8, sorry, Sonet 5 versus GLM 5.2. And for me personally, I would go with GLM 5.2 when you consider everything, including the costs and the fact that you can plug GM 5.2 into Hermes, but you can't do that with something like Sonic 5. So, we've got a question here from Mah who's, you know, thinking about starting a side gig, helping local restaurants, etc. get found online. So,

it sounds like you're basically looking for help with SEO there. The way that I would recommend doing that is, you know, in terms of a stack, probably go with just Claude. If you want the full basics, just go with Claude and then use a dress too. If you want to go a bit more advanced, then we actually have a full SEO system inside the agent operating system. So, for example, it can pull in the latest research

from your Google search console with a Google Workspace API. You've also got Open SEO which works with the data for SEO API for rank trackings inside AI and also inside Google. And then we also have this system where we can plug in a keyword, plug in a case study, deploy that to our websites with Netleify or WordPress and from there actually index the content with indexal as well. So if you want like a full stack built

out, you could use that. If you want the basic stack, you've got that above too. Q was asking about like cheaper Frontier models. So for us, we've actually used Olama with GLM 5.2. That's pretty good. You can get the cloud subscription plan that also gives you access to loads of other models, too. And then what you can do is you can actually set up in one click GM 5.2 with claude code like we've got built right

here or you can actually set up Hermes in one single click with GM 5.2 as we've got inside the agent OS as well. So that's a good option. Otherwise just a normal GLM 5.2 coding plan is pretty good too. You also might think, you know, some of this stuff is a bit complicated or that sort of thing. So, you can see, for example, like Teemo, he just joined like literally he's brand new and he's already set

up Hermes agent and set up his first Hermes system. So, if you're new to AI agents, like there's so many people inside the community who are getting awesome results. There's 194 pages of wins and testimonials inside our um document as you can see right here. So I know like if I'm nontechnical and I can set this up, if they're all nontechnical and they can set this up, anyone can win with AI. You just have to learn

how. And I think a community that's supportive and positive like this is great for learning that stuff. So shout out to Teemo, but also use this as inspiration. If you're struggling with AI agents or you're struggling with using AI, you can get there. You just have to be persistent and it's a skill that you learn over time. So, we got a win from John Meza, who is an absolute legend inside the community. And you can see

what he's built here. So, he's got Hermes agent running directly with his agent OS over here, and he's plugged it into the system, which is quite a nice way to use it. Now we've got Hermes as well with a bit more advanced functionality inside our agent operating system. But the main point is if you're using anything open source like Hermes or openclaw definitely use it like this. Now one way that I like to get the most

out of Hermes as well is to basically segment it by what it can do into a separate tab. So what I mean by this for example is like we've got a system here where we can do lead generation in one prompt. Then we can draft the email campaign and then we can manage our emails and all the leads that we've contacted inside this one single system. And that's just one tiny piece of what Hermes can do.

But we've got it inside one workflow so we can come back to it whenever we need to. We also for example got the video agent over here and we can generate videos in one click with Hermes just by typing in the prompt and then for example we can generate the full avatar video as you can see. So however you prefer to use Hermes inside your agent operating system you can. This is how I recommend using it.

Um it's a great win for John as well. Shout out to John for taking action and also sharing that with the community. I've got a question from Sierra, which is how to get Hermes and Alpha working. So, our alpha was a free API that was available on beta and it was released end of April. I actually saw recently that it's now been taken off and it's not free anymore. So, the model our alpha doesn't exist on

open route anymore. If you actually type in our alpha, the only model that pops up is Mansa Weaver and that is a paid API. But what you can do instead is if you type in free inside open router There's a bunch of free models that you can use instead. So for example, Gemma 4, North Mini Code, Hermes 3, Nvidia Neatron 3.5, etc. So that's a good way around it instead. This pretty interesting. So Robert was asking

like are there any themes for the agent OS which is not something we've actually done. It might be something we create in the future if there's enough demand for it. But right now we've just got the UI as is. However, what I will say is that when I look at a lot of community members who are using the agent OS, a lot of community members using the agent OS, they actually customize it how they want it

to look. And the great thing about that is you can always change the branding, the logo, you can change the style of it, look and theme. Sheena did a great job of this and actually rebranded it plus changed it to her own version. So you can customize yours as much you want. There was a good example from Robbie D as well who created his own sort of canban pipeline for job outreach with the agent OS as

well. So whatever theme style color etc you prefer to use you can set up with your own agent OS. There was also a good example of Joseph Hston who basically changed the colors inside his Obsidian memory setup. So it's very easy to customize and change however you want. You can just ask the agent that set out for you to do that for you. So we have a question from Nova who was asking about what you know

what's the best system to use for memory. So we currently use Obsidian which is just a bunch of markdown files but Nova is using SQL and he said somehow I feel a bit of pressure to instead of use Obsidian use SQL and he asked like is Obsidian somehow better is SQL better or what should you use for memory? I personally don't use SQL just because it's a bit more technical and it's based on code whereas for

example with markdown it's simplified and for me I always find like the simplest systems all things being equal tends to be the best and also when I've looked at other people who have set up a great memory system the best ones I've seen by far have been Obsidian I've never seen like a really good memory system created with SQL so if you want to simplify and if you want to make things easy for yourself and your

agents, then I would go with Obsidian. But for me personally, I don't have a lot of experience with coding with SQL, so I can't recommend it. So, Teemo was asking about whether to use task MD and create markdown tasks for Kamban or how do you orchestrate your agents? For me personally, I don't use Taskm. I use Hermes with Kamban instead. Uh, Taskm is not something I'm really familiar or something I've tested with personally. If you do

test out Taskm, just let us know cuz I'd be interested to know, okay, which one do you prefer? But for me right now, just running Hermes agent and then that drives our camb board forward seems to be the fastest and easiest solution. And we've got some great outputs from the camb. They actually just released a new update, the Hermes cam yesterday as well, so it doesn't freeze as much. But yeah, Taskm is not something I'm familiar

with. I tend to just go with the simplest systems because otherwise if you make things too complicated for reagents, they just kind of break or there's more problems in the system. So if you can simplify as well, that tends to work better. Tom was asking about Lama versus Hermes. Which one is best? So, it's kind of like comparing apples and oranges. Olama is a way to run models and you can run models locally or via the

cloud with Olama and they actually have an option where you can run Hermes in one click with Olama. And then for example with Hermes, Hermes is the agent, right? So you want to think of it in terms of okay, so Olama is the model provider and Hermes is the agent that you can plug the model into. So Hermes is for example the hands and alarm is a brain and they work together. It's not something you would

compare side by side. Got another question here from Teemo. So Teemo was asking, you know, what is the best memory system? Should you use Bite Rover? Should you use Open Viking? Should you use Honcho? What should you use? Particularly when you have multiple Hermes profiles. So I would stick with Obsidian. That's what's worked the best for me right now. and we have the memory galaxy plugged into the agent OS using that. The reason that I don't

use the native memory systems with Hermes is like number one they're not some of them are not free and also number two Obsidian seems to work really well with AI agents and it also transfers across clawed openclaw Hermes and whatever else comes out next. So for me personally, I just use Obsidian. And the great thing about it is it's a very simple system with just markdown files and folders. So it's easy for your agents to update,

navigate, and organize for you. Also, here's another good example. So, you can see here that Adrien, they created a LinkedIn desk inside their agent operating system. So, you can see, for example, how it works. It's using our system, but basically they've added their own custom setup inside the agent operating system. So, it's called the LinkedIn desk, and then it can create content. You can schedule it. You can create the content with AI and it's all inside

one tidy workflow that's one click away inside the system. So that is an amazing example of what you can build with the agent OS and how you can customize it and make it your own. They're also asking for some help which Robbie has helped with already. So appreciate that Robbie. But yeah, the the beauty of the agent operating system is if you have a workflow, you can set it up so that it's only one click away.

So if we need to do videos, we've got over here. If we need to do SEO, we've got that over here. If we need to check the memory, we've got that over here. If we need to run a team of agents inside a cam board, we've got over here. And it's all set up and ready to go whenever you need it. And I think that's the beautiful thing about this is it's very customizable. And also, you've

seen how many members are just taking these systems and making it their own. So, we have answered every single question inside the AR prof. If you want to ask me questions like this, I create a video tutorial based on all the latest questions. We've got an amazing community that's really supportive, helpful, lots of great people inside there, lots of people helping each other. 194 pages of testimonials and wins from people winning with AI. It's just a

great community to be part of as well. And I think, you know, when you're learning AI, it can be quite a lonely journey. You know, you're you're coding all the time. you're testing out new tools. It's very hard to meet people who are into AI as much as you are, I think, um because it's so new. So, if you want to join a community where you can meet people and network with great entrepreneurs, founders, people using

AI agents plus myself, then you can do that inside the AirPuff. You also get the full agent operating system that I've shown you today with every system plugged in already. And then inside the community, you can ask questions and get help and video tutorials. I personally answer every single question. We also showcase like what members are doing. So you can see for example this system that Robbie set up inside the classroom you get access to all

of our best trainings. So we have for example this section with the agent OS system the last update date a zip file to install it and then we have new daily tutorials based on what's actually useful with full step-by-step guides and videos. And then additionally if you're a complete beginner you can go from beginner to expert over here. You can also get weekly coaching calls. So you can get four weekly coaching calls, get help and support

in real time. Inside the map, you can meet people in your local area who are working on stuff like this. So I hope to see you inside there. Link in the comments description or go to the aiprofitborn.com. See what questions we got here. Wow. Okay. Uh that's kind of what blah blah blah. How do you use OMI with Obsidian? Yeah. So, OMI uh basically the way it work is like OMI can um it can record your

screen, it can listen to your conversations and then it takes notes on your memories and then what you can do is export all the notes that it takes daily over to Obsidian and that's how they sync together because OMI is taking notes and creating the memory and then Obsidian stores the memory so that your agents can use that context. So, that's how they work together. Do you ever think Google Gemini will have the top model in

the world? I think, you know, once a year when they release something big, they'll have the top model and then give it a month and another model comes out. So, like Google, they don't release that many updates, but when they do, they're really good. So, it's usually a case that they create something good, then it drops, then they get overtaken again, and it kind of follows that cycle. So, it's one of those things, isn't it? Where

it's up and down like a yo-yo. Is OMI Mac only? Yes, but we have a tutorial on a different tool you can actually use over here if you don't have a Mac. And this is another open source uh system you can use, which is this one, as you can see. So, there's a full video tutorial and step-by-step guide on how to use that here if you prefer to use something like that instead. Hey hey hey. Heat.

Heat. N. Daddy, Hey, hey, hey. Heat. Heat. Daddy, daddy. Heat. Heat. N. Ryan says, "Can you loop with Sonic 5?" Yes, you can. We actually have a loop engine inside the agent OS here. So you can see the loop section. So you could switch this to whatever API you wanted. I mean even like pretty much any model you could loop, you know, like it's it's just having the right system that can do the looping for you.

But yeah, for sure with Sonic 5, brand new API, probably very effective at that to be fair, but not as effective as Opus 4.8. So I would just stick to Opus 4.8. Today we have a brand new update from a Chinese model that's open source. It is called Long Cat. Long Cat 2.0 is here. And this was actually the full model behind our alpha. So if you're familiar with our alpha which was a free API you

could actually use it with Hermes. You could plug it into claude code before and it was not bad at all. Is a genetic model and this was actually revealed as longat 2.0. So this is now officially been released and you can get access to it and you can see the full details right here. here. So, it's got long cat, sparse attention, um, zero compute experts MOPD stacks up not badly on the benchmarks here if you're wondering

how to compare against everything. So, it's terminal bench 2.1. It holds its own with these other models as you can see right here. This is compared against Opus 4.6, 4.7, and 4.8. Now, obviously, Opus 4.8 is crushing it by a long way as you can see here. Um, the other thing that I noticed is if you're using this directly on the website, so if you go to longcat and then you go to the API section, it

looks like you can't use the API unless you have some sort of Chinese setup. So for example here if we try and get a pack it's broken anyway. Look at that. Um but if if you want to use it I've been testing out. So I'll show you what we've done so far with it and how it works and how it performs here. So, let's have a look at some of the stuff we built with this. How

it performs some benchmarks. Some of it was good, some of it was not so good. So, this is a game we created called Dragon Realm. It's It's not bad. I mean, the graphics are not bad. This is nowhere near the same standard as GM 5.2 or Opus 4.8 or anything that I don't think it's designed to be. I think it's just designed to be a cool model that you can test out. This one was not bad

at all as well. So, this is kind of like a a Skyrim style open world game. As you can see, it runs pretty smoothly. The graphics are not great. And then here's another one. This is probably the best output that I saw so far. And we can compare it on Goldiebench against a bunch of other examples in a second. Um, but again, it's still pretty basic and buggy, right? Like, like what is going on here? It

just goes completely black. And then we have another example right here. So overall on my tests, I wouldn't say it's that impressive, we ran it through Goldie Bench. I'll show you how it compares versus our models in a second. It is a 1.6 open source 1.6 trillion parameter model. And the other thing to note here, this is probably the biggest update about it as you can see from this tweet by Robin is that number one it's

open source but number two was actually built on a different chip. So this was built on China's matrien and this was trained without a single Nvidia chip. Now, if you're wondering who are Matri, they're basically like China's version of Door Dash, which is pretty crazy. I mean, we saw this with Xiaomi as well earlier this year where like, you know, everyone's getting involved in AI. Everyone's bringing out their own models. Companies that we don't expect to

bring out the models come out. But yeah, it's pretty cool. I mean, like, fair play to them for bringing out a model like this. Would I say it's the best model I've ever used? Probably not. But it is fun to play with, fun to test out etc. The way that I actually used it because I can get access to the API and see like it's it's not even available to top up yet is you can actually

go to the chat section and just start using it there. They've also got a full breakdown of how it works step by step as you can see on the website. Also on the evaluations, this is pretty interesting. GPT 5.5 is only slightly above on terminal bench versus longat. And then on swbench pro, longat is actually outperforming GP5.5. Now again, I've shown you my tests. Do I personally think that it's better than GBT 5.5 on my own

benchmarks? Probably not. And Opus 4.8 is is beating them all, but you would expect that anyway. Now, I think probably the most relevant sideby-side comparison is comparing it against GLM 5.2 because that is another opensource project that just came out of China recently and is very cheap on the coding plan and also it's it's designed to be agentic as well. So, for example, if we have a look at this cry game, this is the one created

from GLM 5.2. It's pretty dark and it is a little bit buggy. But it works and it's actually, you know, interesting and quite useful. If we compare that versus the option from Lancat here, you can see that this is very limited. It has nothing going on in the game and it's super buggy, right? You can just walk through the wall. So on that particular benchmark, GLM 5.2 is winning. Let's have a look at the next one.

So this is the dragon rail game as we showed earlier. Now this is the one from GLM 512. Pretty nice open world, has lots of interesting stuff, quite playable, etc. Right. Um and the even like the graphics are not bad at all. Now, if we compare that versus this option from Lancat here, you can see that it's just not quite the same quality or caliber. It kind of feels like this is an older generation of it.

Let's check the next version now. So, we've got our own version of Skyrim over here. This is the GLM 5.2 option. Again, looks pretty nice. Very open world. Feels expansive and big. If we compare that versus long catat here, this is actually a lot smoother, but there's just not much going on, right? There's it's not as interesting. And then we created Voxil Craft as well. So, this is the option from GM 5.2, which is actually pretty

impressive. Like, it's it works perfectly. It's fun to play, it's easy to set up, etc. If we compare that versus the version from Lancat, you can see it's just nowhere near the same level. So, if you have a choice between them, I would still go with GLM 5.2. I think that it's interesting to see what's coming out of China, but I think GLM 5.2 by far is the strongest model out of everything I've tested. when it

comes to open source stuff. So that's basically it. Will I be switching to longat anytime soon? No. But I wanted to test it, show you what it looks like, etc. One thing we've actually done with GM 5.2 recently is plug it into claude code inside the agent operating system. And this is our mission control dashboard where we can have all of our agents working and plugged in together. Now, if you want to get that set up,

it's all available inside the AI profit boardroom. Link in the comments description or go to the profitboard.com. Inside the community, you can ask questions, get help and support real time. I answer every single question inside the community personally with a video tutorial. Plus, you get the whole support and help of the whole community. And there's always someone online 24/7 to help you. Inside the classroom, you can get access to all of our best trainings and lessons.

Inside the new daily update section, you can see we've got our full agent OS system, the last update date. So, we update this daily video tutorial and a zip file on how to use it. And then we add new tutorials based on what actually comes out and is useful. Inside the calendar, you can jump on weekly coaching calls, get open support in real time. Inside the map, you can meet people in your local area who are

building with AI agents like you've seen today. Um, that's all available inside a profit boarding link in the comments description or go to the apiprofit.com. Thanks for watching. Is anthropic API? Yeah. So, you can't use OAF. You cannot use OAF from Claude with Hermes agent. Do you think anthropic will release tomorrow with Fable 5? Yeah, 100%. They've already uh confirmed that they will. So, I think there'll be an announcement in a few hours and then I

would expect to see Fable 5 coming back for some people within the next 24 hours with the full roll out probably coming within the next sort of 48 hours as well. Uh I don't think I think it'll be part of the subscription actually. Um, obviously the tokens will be limited, but I think Fable 5 will be part of the subscription from what I've seen announced as well recently. So, that's pretty exciting, too. Heat. Heat. Team rocker

is here. Welcome. Will it be? Yeah, I would expect that it will still be included inside the subscription at least for now. Now, you may not be able to use it that much, you know, in terms of the token usage, but I would expect that yeah, you you won't need to pay for the API at first at least. Anyway, do I use GLM 5.2? Yes, we have it plugged into our agent operating system and we use

it inside uh we use GLM 5.2 inside GLM code over here and then we can create and build whatever we want with it. And then we also have GLM 5.2 as a CLI and then the other way that I use it is we plug it into Hermes agent and we've got GM 5.2 as one of the agent profiles. So we can build and automate what we want with GM 5.2. I also tested it on Goldiebench with

42 different builds. So, if you want to see what we've built with it, what it can do, what sort of stuff you can create with it, just check out the full systems here and all the builds. And you can see how it performs versus everything else on the compare pages here, too. Uh, you can just go to Z AI. So, if you go to Z AI, you can use GLM 5.2 for free. What's your thoughts on

Sonic 5? Not going to use it. Not going to use it. I've tested out um you can see it on Goldie Bench if you want to see like the benchmarks and everything. So, we've already built out a bunch of stuff with it and compared it with several models. So, you can see on the leaderboard down here, number 10, but from what I'm seeing so far, it's it's nowhere near the same level as uh you know, Fable

5. Of course, it's not a big step up from Sonic 4.6, honestly. And if you look at Sonic 5 and you compared it side by side versus GLM 512 or Opus 4.8, I would go with Opus 4.8. I think that announcement will come in a few hours. So I would give it like 6 to 12 hours, I can imagine. And then I would imagine that the update will come. And that'll probably come before the full roll

out. Maybe like an hour or two before the Fable 5 model starts coming back to everyone. back in a minute. Your ranking system is subjective. Every ranking system is subjective, my friend. Um, we've all seen that, you know, we've all seen systems where we see the benchmarks and then we see what actually happens in reality. you know, whether you see numbers or not, benchmarks, they tend to be subjective and they don't tend to live up to

what you actually see. So, my philosophy is always test it yourself. You can see what I've built with it, get a feel for it yourself, and then you can test it yourself and see what you think as well. I think that's the only way right way to use and test AI, right? If you don't test it yourself, um, then you're just kind of guessing or you're relying on someone else's word. But bear in mind like most

of these benchmarks as well they come out from the company directly. So do you trust your benchmarks or someone who's neutral or do you trust their benchmarks? That's your choice, isn't it? Hey, hey, hey. Heat. Heat. Daddy, daddy. Hey. Hey. Hey. Wait. Yeah, I agree with that. I think front end is better. Oh, announcement just came out. Let's have a cheeky gander. Thank you, sir, for telling me. Anthropic or Claude. Let's have a look at both.

It's not on Claude's account. I mean, it's not on uh Anthropit. Where's the announcement? Oh, and the anthropic news site. Let's have a look. Thanks for the heads up, by the way. Can't see it. Where's the link? Redeployment. All right, perfect. There we go. Boom shack lacquer. Thank you very much for letting me know, by the way. I appreciate that. So, it looks like we're back in the game, my friends. So, Anthropic have announced it. They

they announced it on their Twitter account earlier today and then they said there's going to be a big announcement on the website and here it is. So, June the 12th, export control happens. Now, it's been lifted. Happy days. It's going to be available starting tomorrow, Wednesday, July the 1st. to users globally on claude and everywhere else for pro max team and select enterprise plans. Now, Fable 5 will be included for up to 50% of weekly usage

limits through July the 7th, after which it'll be available via usage credits. And they're going to reenable access on AWS, Google Cloud, and everywhere else as quickly as possible. So, I guess it's probably 10:30 EST in Claude in um the US right now. So, 10:30 to p.m. on June the 30th. So, I I can imagine within 12 hours we should start seeing access to Fable 5 again, which is great news. They've also restored access to Mythos

5 for a set of US organizations. We all knew we weren't getting access to that anyway. And then they're going to be a there's a full full full guide on here. All right. Nice. I'm going to do a separate tutorial on everything else that is inside this announcement, but pretty interesting stuff. Awesome. Awesome news. Thank you so much for giving me the heads up. I appreciate it. Thanks everyone who joined. DJ Jerome, you're an absolute legend.

Thanks to everyone else who joined as well today. I'll see you on the next one, peeps. Cheers. Bye-bye.