Not even close… M5 Max vs Razer Blade 18

Last year, I put the Razer Blade 18 up Last year, I put the Razer Blade 18 up against my MacBook, and it was not even against my MacBook, and it was not even against my MacBook, and it was not even close. Even the M4 Pro beat it. Come on. close. Even the M4 Pro beat it. Come on. close. Even the M4 Pro beat it. Come on. Well, this is the new one, and this got Well, this is the new one, and this got Well, this is the new one, and this got the RTX 5090 instead of the 4090, 24 the RTX 5090 instead of the 4090, 24 the RTX 5090 instead of the 4090, 24 gigs of VRAM instead of 16, and a brand gigs of VRAM instead of 16, and a brand gigs of VRAM instead of 16, and a brand new Core Ultra 9 from Intel inside. And new Core Ultra 9 from Intel inside. And new Core Ultra 9 from Intel inside. And even the stickers have been updated, so even the stickers have been updated, so even the stickers have been updated, so it's got to be better, right? Well, it's got to be better, right? Well, it's got to be better, right? Well, that's what I want to find out. Because that's what I want to find out. Because that's what I want to find out. Because last year, my complaint wasn't how it last year, my complaint wasn't how it last year, my complaint wasn't how it was slow. It was more about what you got was slow. It was more about what you got was slow. It was more about what you got for your money because these machines for your money because these machines for your money because these machines are expensive. How can Razer charge this are expensive. How can Razer charge this are expensive. How can Razer charge this much money? I don't know. Now, I don't much money? I don't know. Now, I don't much money? I don't know. Now, I don't want to get too much into synthetic want to get too much into synthetic want to get too much into synthetic benchmarks because they don't really benchmarks because they don't really benchmarks because they don't really mean much, but they are a good indicator mean much, but they are a good indicator mean much, but they are a good indicator of where things stand. So, I'm going to of where things stand. So, I'm going to of where things stand. So, I'm going to run Geekbench 7 this time. And right out run Geekbench 7 this time. And right out run Geekbench 7 this time. And right out of the box, the Razer comes in balance of the box, the Razer comes in balance of the box, the Razer comes in balance mode and it's also unplugged. By the mode and it's also unplugged. By the mode and it's also unplugged. By the way, Geekbench maps to stuff you really way, Geekbench maps to stuff you really way, Geekbench maps to stuff you really feel all day long. For developers, feel all day long. For developers, feel all day long. For developers, single core would be like your editor, single core would be like your editor, single core would be like your editor, that's the typical latency. your that's the typical latency. your that's the typical latency. your language server. That's TypeScript language server. That's TypeScript language server. That's TypeScript checking your types and that's whether checking your types and that's whether checking your types and that's whether the machine feels snappy or feels like the machine feels snappy or feels like the machine feels snappy or feels like it's fighting you. And the multi-core it's fighting you. And the multi-core it's fighting you. And the multi-core score is everything you wait on like score is everything you wait on like score is everything you wait on like builds and compiles, running test builds and compiles, running test builds and compiles, running test suites, Docker spinning up, your suites, Docker spinning up, your suites, Docker spinning up, your bundling for production. And the thing bundling for production. And the thing bundling for production. And the thing that matters a lot for a laptop that matters a lot for a laptop that matters a lot for a laptop specifically is you're going to be specifically is you're going to be specifically is you're going to be taking it places. So, you're not always taking it places. So, you're not always taking it places. So, you're not always going to be plugged in. Now, this is the going to be plugged in. Now, this is the going to be plugged in. Now, this is the Apple MacBook M5 Max. And the energy Apple MacBook M5 Max. And the energy Apple MacBook M5 Max. And the energy mode usually comes on automatic. So, it mode usually comes on automatic. So, it mode usually comes on automatic. So, it kind of allocates the resources where it kind of allocates the resources where it kind of allocates the resources where it needs them. I did set it to high power.

needs them. I did set it to high power. needs them. I did set it to high power. So, let's run that. I know you're going So, let's run that. I know you're going So, let's run that. I know you're going to say, "Oh, Alex, it's not fair. This to say, "Oh, Alex, it's not fair. This to say, "Oh, Alex, it's not fair. This one is balanced and that one's high one is balanced and that one's high one is balanced and that one's high power." Don't worry, we'll run this high power." Don't worry, we'll run this high power." Don't worry, we'll run this high power. Also, there's just a couple of power. Also, there's just a couple of power. Also, there's just a couple of different settings you need to tweak different settings you need to tweak different settings you need to tweak here. Also, while it's running, just to here. Also, while it's running, just to here. Also, while it's running, just to address some of the comments from last address some of the comments from last address some of the comments from last year, yeah, this is a gaming laptop and year, yeah, this is a gaming laptop and year, yeah, this is a gaming laptop and this is a productivity laptop, but you this is a productivity laptop, but you this is a productivity laptop, but you can also game on a Mac and you can game can also game on a Mac and you can game can also game on a Mac and you can game on this. The Razer will give you a much on this. The Razer will give you a much on this. The Razer will give you a much better gaming experience. And also, most better gaming experience. And also, most better gaming experience. And also, most people I know are not going to buy two people I know are not going to buy two people I know are not going to buy two $5,000 to $6,000 laptops. They're going $5,000 to $6,000 laptops. They're going $5,000 to $6,000 laptops. They're going to buy one. So, if this is the one to buy one. So, if this is the one to buy one. So, if this is the one you're buying, you're doing your you're buying, you're doing your you're buying, you're doing your productivity stuff on it and you're productivity stuff on it and you're productivity stuff on it and you're gaming on it. And you want to know how gaming on it. And you want to know how gaming on it. And you want to know how it behaves with productivity, too, to it behaves with productivity, too, to it behaves with productivity, too, to see what you're getting or what you're see what you're getting or what you're see what you're getting or what you're not getting. Here's what you're getting not getting. Here's what you're getting not getting. Here's what you're getting unplugged. Single core score 1,840, unplugged. Single core score 1,840, unplugged. Single core score 1,840, multi-core 15,118. multi-core 15,118. multi-core 15,118. It's not amazing, which is why for the It's not amazing, which is why for the It's not amazing, which is why for the rest of this video, I'm actually going rest of this video, I'm actually going rest of this video, I'm actually going to have it plugged in. Now, with the to have it plugged in. Now, with the to have it plugged in. Now, with the Mac, it doesn't matter if it's plugged Mac, it doesn't matter if it's plugged Mac, it doesn't matter if it's plugged in or not. It's constantly using full in or not. It's constantly using full in or not. It's constantly using full power like I set it to. But with this power like I set it to. But with this power like I set it to. But with this one, I think we can get an improvement.

one, I think we can get an improvement. one, I think we can get an improvement. I plugged it in, and I'm also going to I plugged it in, and I'm also going to I plugged it in, and I'm also going to set this to high performance. There's set this to high performance. There's set this to high performance. There's two places where you can actually do two places where you can actually do two places where you can actually do this. One is right here in settings this. One is right here in settings this. One is right here in settings under system power, and then you can set under system power, and then you can set under system power, and then you can set this to plugged in best performance on this to plugged in best performance on this to plugged in best performance on battery, best performance, or you can battery, best performance, or you can battery, best performance, or you can pick and choose what you need. But pick and choose what you need. But pick and choose what you need. But there's also another way to get high there's also another way to get high there's also another way to get high performance power plan, not power mode, performance power plan, not power mode, performance power plan, not power mode, and that's in the control panel under and that's in the control panel under and that's in the control panel under power options. Right now, it's set to power options. Right now, it's set to power options. Right now, it's set to balanced, and it doesn't give you balanced, and it doesn't give you balanced, and it doesn't give you another option to set, but you can do it another option to set, but you can do it another option to set, but you can do it through the command line. Just go to the through the command line. Just go to the through the command line. Just go to the terminal as admin, and you can run power terminal as admin, and you can run power terminal as admin, and you can run power cfg, power config, duplicate scheme. You cfg, power config, duplicate scheme. You cfg, power config, duplicate scheme. You have a long grid. I'll put that grid have a long grid. I'll put that grid have a long grid. I'll put that grid down below for high power mode. You down below for high power mode. You down below for high power mode. You basically duplicate that scheme. In this basically duplicate that scheme. In this basically duplicate that scheme. In this case, I duplicated the ultimate case, I duplicated the ultimate case, I duplicated the ultimate performance scheme, but it's going to be performance scheme, but it's going to be performance scheme, but it's going to be high power. And then power config set high power. And then power config set high power. And then power config set scheme min. Boom. You have to back out scheme min. Boom. You have to back out scheme min. Boom. You have to back out and then go back in. And now you have and then go back in. And now you have and then go back in. And now you have high performance magically there. And high performance magically there. And high performance magically there. And select it. It says favors performance select it. It says favors performance select it. It says favors performance but may use more energy. Now I have to but may use more energy. Now I have to but may use more energy. Now I have to say that last year when I plugged the say that last year when I plugged the say that last year when I plugged the Razer in, it started spinning up its Razer in, it started spinning up its Razer in, it started spinning up its fans going crazy. This year this is a fans going crazy. This year this is a fans going crazy. This year this is a very quiet machine. I'm actually pretty very quiet machine. I'm actually pretty very quiet machine. I'm actually pretty impressed. Also, when you set high impressed. Also, when you set high impressed. Also, when you set high performance that way, power mode is no performance that way, power mode is no performance that way, power mode is no longer selectable. So, it's going to be longer selectable. So, it's going to be longer selectable. So, it's going to be set to best performance all the time, set to best performance all the time, set to best performance all the time, which is the way I like it. I'm going to which is the way I like it. I'm going to which is the way I like it. I'm going to run Geekbench one more time. It run Geekbench one more time. It run Geekbench one more time. It finished. And I got to tell you finished. And I got to tell you finished. And I got to tell you something here because I actually ran it something here because I actually ran it something here because I actually ran it two different ways. And this is a big two different ways. And this is a big two different ways. And this is a big change from last time. One way is under change from last time. One way is under change from last time. One way is under high performance just like I showed you.

high performance just like I showed you. high performance just like I showed you. These are the scores. Single core 2,918. These are the scores. Single core 2,918. These are the scores. Single core 2,918. Multi-core 24,149. Multi-core 24,149. Multi-core 24,149. Really good. I also ran it under Really good. I also ran it under Really good. I also ran it under balanced again and that gave me 29,08 balanced again and that gave me 29,08 balanced again and that gave me 29,08 23,957. 23,957. 23,957. essentially the same scores, high essentially the same scores, high essentially the same scores, high performance or balanced. And what performance or balanced. And what performance or balanced. And what matters more here is that it's plugged matters more here is that it's plugged matters more here is that it's plugged in or not plugged in. That's what makes in or not plugged in. That's what makes in or not plugged in. That's what makes the big difference. So, let's take the the big difference. So, let's take the the big difference. So, let's take the higher of the two, which is this one, higher of the two, which is this one, higher of the two, which is this one, the high performance one, slightly the high performance one, slightly the high performance one, slightly higher. And compare it to the Mac. Oh, higher. And compare it to the Mac. Oh, higher. And compare it to the Mac. Oh, uh, wait a minute. 3,220 on single core uh, wait a minute. 3,220 on single core uh, wait a minute. 3,220 on single core and 22,658 and 22,658 and 22,658 on multi-core. Well, that's not actually on multi-core. Well, that's not actually on multi-core. Well, that's not actually true. That's inside a Windows 11 virtual true. That's inside a Windows 11 virtual true. That's inside a Windows 11 virtual machine running inside the Mac. So, we machine running inside the Mac. So, we machine running inside the Mac. So, we got 24 cores here on the Razer machine got 24 cores here on the Razer machine got 24 cores here on the Razer machine and 12 cores in a virtual machine on top and 12 cores in a virtual machine on top and 12 cores in a virtual machine on top of Mac OS and it's within 6% of the real of Mac OS and it's within 6% of the real of Mac OS and it's within 6% of the real thing running on bare metal with 24 thing running on bare metal with 24 thing running on bare metal with 24 cores. I'll do more virtual machine cores. I'll do more virtual machine cores. I'll do more virtual machine stuff a little bit later to do some stuff a little bit later to do some stuff a little bit later to do some Visual Studio tests, but here's the Visual Studio tests, but here's the Visual Studio tests, but here's the actual score on the Mac itself. 3756 actual score on the Mac itself. 3756 actual score on the Mac itself. 3756 for single core, 35,623 for single core, 35,623 for single core, 35,623 on multicore. That is crazy. I don't on multicore. That is crazy. I don't on multicore. That is crazy. I don't expect one AI prompt to build an entire expect one AI prompt to build an entire expect one AI prompt to build an entire project for me. In reality, I'm project for me. In reality, I'm project for me. In reality, I'm constantly moving between models constantly moving between models constantly moving between models depending on the job. GPT for research, depending on the job. GPT for research, depending on the job. GPT for research, Claude for coding, Gemini for massive Claude for coding, Gemini for massive Claude for coding, Gemini for massive context, Nano Banana, Midjourney, Flux context, Nano Banana, Midjourney, Flux context, Nano Banana, Midjourney, Flux for images, and then I've got Seed Dance for images, and then I've got Seed Dance for images, and then I've got Seed Dance and Cling for video. That's why Chat LM and Cling for video. That's why Chat LM and Cling for video. That's why Chat LM by Abacus AI makes sense. It brings day by Abacus AI makes sense. It brings day by Abacus AI makes sense. It brings day one support for the latest GPT, Claw,

one support for the latest GPT, Claw, one support for the latest GPT, Claw, Gemini, Grock, DeepSeek, and more in one Gemini, Grock, DeepSeek, and more in one Gemini, Grock, DeepSeek, and more in one place the moment they drop. Pick any place the moment they drop. Pick any place the moment they drop. Pick any model from the interface or let route model from the interface or let route model from the interface or let route LLM automatically choose the best model LLM automatically choose the best model LLM automatically choose the best model for each prompt. Create professional for each prompt. Create professional for each prompt. Create professional presentations with graphs and charts and presentations with graphs and charts and presentations with graphs and charts and deep research detailed content. Need deep research detailed content. Need deep research detailed content. Need human sounding copy? Human eyes rewrites human sounding copy? Human eyes rewrites human sounding copy? Human eyes rewrites text to defeat AI detectors. Need text to defeat AI detectors. Need text to defeat AI detectors. Need visuals? Pick frontier or open-source visuals? Pick frontier or open-source visuals? Pick frontier or open-source models. And when you need more than models. And when you need more than models. And when you need more than chat, Abacus AI agent can help build chat, Abacus AI agent can help build chat, Abacus AI agent can help build complex apps and websites, connect complex apps and websites, connect complex apps and websites, connect payments, or run 24/7 agents that keep payments, or run 24/7 agents that keep payments, or run 24/7 agents that keep working through longer tasks. The best working through longer tasks. The best working through longer tasks. The best part is app hosting, back-end database, part is app hosting, back-end database, part is app hosting, back-end database, and O support comes with the and O support comes with the and O support comes with the subscription. All that starts at just subscription. All that starts at just subscription. All that starts at just $10 a month. Way cheaper than paying for $10 a month. Way cheaper than paying for $10 a month. Way cheaper than paying for all those subscriptions separately. all those subscriptions separately. all those subscriptions separately. Check out chatlm.abacus.ai Check out chatlm.abacus.ai Check out chatlm.abacus.ai or click the link below. Moving on. Now, or click the link below. Moving on. Now, or click the link below. Moving on. Now, this one runs right in the browser. this one runs right in the browser. this one runs right in the browser. Boom. Now, I know that some of you are Boom. Now, I know that some of you are Boom. Now, I know that some of you are thinking, "I don't do web development. thinking, "I don't do web development. thinking, "I don't do web development. Why should I care?" Well, here's why. Why should I care?" Well, here's why. Why should I care?" Well, here's why. This test speedometer runs real This test speedometer runs real This test speedometer runs real framework loads, React View, Angular, framework loads, React View, Angular, framework loads, React View, Angular, the stuff a lot of us actually ship. I'm the stuff a lot of us actually ship. I'm the stuff a lot of us actually ship. I'm a web developer. That's the kind of a web developer. That's the kind of a web developer. That's the kind of stuff I work on. But here's the part stuff I work on. But here's the part stuff I work on. But here's the part that people miss. If you're running VS that people miss. If you're running VS that people miss. If you're running VS Code or Cursor or Windsurf, even if Code or Cursor or Windsurf, even if Code or Cursor or Windsurf, even if you're not doing web development, you're you're not doing web development, you're you're not doing web development, you're already using the browser. That's already using the browser. That's already using the browser. That's Electron. So, this score isn't just Electron. So, this score isn't just Electron. So, this score isn't just websites. It's your editor. It's your websites. It's your editor. It's your websites. It's your editor. It's your dev server, your dev tools with 30 tabs dev server, your dev tools with 30 tabs dev server, your dev tools with 30 tabs open. tab switching between them open. tab switching between them open. tab switching between them quickly. That's most of your day. 48.3.

quickly. That's most of your day. 48.3. quickly. That's most of your day. 48.3. That's actually the best score I've seen That's actually the best score I've seen That's actually the best score I've seen on a Windows laptop. That's really good. on a Windows laptop. That's really good. on a Windows laptop. That's really good. Last year's machine got 23. On the Mac, Last year's machine got 23. On the Mac, Last year's machine got 23. On the Mac, it's 49.4. So, they're basically the it's 49.4. So, they're basically the it's 49.4. So, they're basically the same. That's actually surprising. Oh. same. That's actually surprising. Oh. same. That's actually surprising. Oh. Uh, oops. That's in the virtual machine Uh, oops. That's in the virtual machine Uh, oops. That's in the virtual machine again. again. again. Okay, fine. I won't do that trick Okay, fine. I won't do that trick Okay, fine. I won't do that trick anymore on you. That's more like it. anymore on you. That's more like it. anymore on you. That's more like it. 63.5 on the M5 Max here. It's very hard 63.5 on the M5 Max here. It's very hard 63.5 on the M5 Max here. It's very hard to beat an M5 chip here. Apple has been to beat an M5 chip here. Apple has been to beat an M5 chip here. Apple has been the leader in single core performance the leader in single core performance the leader in single core performance since Apple Silicon came out back in since Apple Silicon came out back in since Apple Silicon came out back in 2020. But I want to be clear here that 2020. But I want to be clear here that 2020. But I want to be clear here that almost 50 is really good for a Windows almost 50 is really good for a Windows almost 50 is really good for a Windows laptop. On a related note, this is the laptop. On a related note, this is the laptop. On a related note, this is the web tooling benchmark from the V8 team. web tooling benchmark from the V8 team. web tooling benchmark from the V8 team. Yeah, that same V8 that created the Yeah, that same V8 that created the Yeah, that same V8 that created the JavaScript engine. And boom, let's go. JavaScript engine. And boom, let's go. JavaScript engine. And boom, let's go. Now, this is your typical JavaScript Now, this is your typical JavaScript Now, this is your typical JavaScript stuff from 8 years ago. the stuff that stuff from 8 years ago. the stuff that stuff from 8 years ago. the stuff that was popular back then, but some of it was popular back then, but some of it was popular back then, but some of it still is. Babel, Typescript, parsers, still is. Babel, Typescript, parsers, still is. Babel, Typescript, parsers, minifiers. There's no disc access and no minifiers. There's no disc access and no minifiers. There's no disc access and no IO here. This is pure stuff here. And IO here. This is pure stuff here. And IO here. This is pure stuff here. And this is the first time I'm hearing the this is the first time I'm hearing the this is the first time I'm hearing the fans of the Razer spin up a little bit.

fans of the Razer spin up a little bit. fans of the Razer spin up a little bit. It's still a very quiet machine. Not It's still a very quiet machine. Not It's still a very quiet machine. Not like last year's, but I'm hearing them. like last year's, but I'm hearing them. like last year's, but I'm hearing them. And we're very close now. Babel 39.6 And we're very close now. Babel 39.6 And we're very close now. Babel 39.6 runs per second on the Mac, 34.4 4 on runs per second on the Mac, 34.4 4 on runs per second on the Mac, 34.4 4 on the Razer. Coffee script 37.2 on the the Razer. Coffee script 37.2 on the the Razer. Coffee script 37.2 on the Mac, 26.06 on the Razer. And Typescript, Mac, 26.06 on the Razer. And Typescript, Mac, 26.06 on the Razer. And Typescript, the one I care about the most, 53.39 on the one I care about the most, 53.39 on the one I care about the most, 53.39 on the Mac, 47.29 on the Razer. Very close. the Mac, 47.29 on the Razer. Very close. the Mac, 47.29 on the Razer. Very close. Geometric mean, 41.9 versus 35.99. This Geometric mean, 41.9 versus 35.99. This Geometric mean, 41.9 versus 35.99. This is actually very good. Did I mention is actually very good. Did I mention is actually very good. Did I mention that I always thought of the Razer as that I always thought of the Razer as that I always thought of the Razer as kind of like the Windows equivalent of a kind of like the Windows equivalent of a kind of like the Windows equivalent of a MacBook? It's solidly built like a tank, MacBook? It's solidly built like a tank, MacBook? It's solidly built like a tank, all metal. And this one being an 18-iner all metal. And this one being an 18-iner all metal. And this one being an 18-iner really feels like what a MacBook would really feels like what a MacBook would really feels like what a MacBook would be if it was 18 in. The keyboard on the be if it was 18 in. The keyboard on the be if it was 18 in. The keyboard on the Razer is also very good and I appreciate Razer is also very good and I appreciate Razer is also very good and I appreciate this little fact. They centered the this little fact. They centered the this little fact. They centered the trackpad even though there is a numpad trackpad even though there is a numpad trackpad even though there is a numpad here. Some laptop makers don't do this. here. Some laptop makers don't do this. here. Some laptop makers don't do this. Anyway, I'm getting a little bit off Anyway, I'm getting a little bit off Anyway, I'm getting a little bit off track here. We're still doing track here. We're still doing track here. We're still doing performance tests, but I'm thinking a performance tests, but I'm thinking a performance tests, but I'm thinking a little bit about hardware and little bit about hardware and little bit about hardware and specifically the SSDs that are inside specifically the SSDs that are inside specifically the SSDs that are inside here because the Mac took a giant leap here because the Mac took a giant leap here because the Mac took a giant leap forward with this new generation because forward with this new generation because forward with this new generation because now they're doing Gen 5 drives. Oh yeah.

now they're doing Gen 5 drives. Oh yeah. now they're doing Gen 5 drives. Oh yeah. Um that's a very different kind of uh Um that's a very different kind of uh Um that's a very different kind of uh drive inside there. It's pretty much two drive inside there. It's pretty much two drive inside there. It's pretty much two times faster than the Razer. Now the times faster than the Razer. Now the times faster than the Razer. Now the Razer is no slouch. Last year's Mac Razer is no slouch. Last year's Mac Razer is no slouch. Last year's Mac model was pretty much the same numbers. model was pretty much the same numbers. model was pretty much the same numbers. We got sequential read speeds of 7,000, We got sequential read speeds of 7,000, We got sequential read speeds of 7,000, 6,300 for writes on the Razer and 13,769 6,300 for writes on the Razer and 13,769 6,300 for writes on the Razer and 13,769 read speed, 13,555 read speed, 13,555 read speed, 13,555 right speed on the Mac. Sequential means right speed on the Mac. Sequential means right speed on the Mac. Sequential means basically you're just taking one large basically you're just taking one large basically you're just taking one large file and just dumping it. But what about file and just dumping it. But what about file and just dumping it. But what about this other number here? The random this other number here? The random this other number here? The random developers care about this one because developers care about this one because developers care about this one because it kind of represents compiled code. And it kind of represents compiled code. And it kind of represents compiled code. And on this one, the Razer does a little bit on this one, the Razer does a little bit on this one, the Razer does a little bit better than the Mac. So, this is going better than the Mac. So, this is going better than the Mac. So, this is going to be important because when you're to be important because when you're to be important because when you're doing compilations, you're not only doing compilations, you're not only doing compilations, you're not only using that processor, you're also doing using that processor, you're also doing using that processor, you're also doing a bunch of disc IO. We'll see that at a bunch of disc IO. We'll see that at a bunch of disc IO. We'll see that at work momentarily. But I also want to see work momentarily. But I also want to see work momentarily. But I also want to see how it does with external read and how it does with external read and how it does with external read and writes since both of these have writes since both of these have writes since both of these have Thunderbolt 5 identical drives inside. I Thunderbolt 5 identical drives inside. I Thunderbolt 5 identical drives inside. I got the 9100 by Samsung, which is the got the 9100 by Samsung, which is the got the 9100 by Samsung, which is the Gen 5 drive. It's one of the fastest Gen 5 drive. It's one of the fastest Gen 5 drive. It's one of the fastest things you can get right now. Of course, things you can get right now. Of course, things you can get right now. Of course, the Mac has three Thunderbolt 5 ports.

the Mac has three Thunderbolt 5 ports. the Mac has three Thunderbolt 5 ports. Razer has just the one. So, let's see Razer has just the one. So, let's see Razer has just the one. So, let's see how we do here. The reason this is how we do here. The reason this is how we do here. The reason this is important is because well when you're important is because well when you're important is because well when you're buying a Mac you are stuck with what you buying a Mac you are stuck with what you buying a Mac you are stuck with what you have. So you kind of tend to buy the have. So you kind of tend to buy the have. So you kind of tend to buy the more expensive models because you might more expensive models because you might more expensive models because you might run out of space or you might just say run out of space or you might just say run out of space or you might just say fine I'm I don't need the really fine I'm I don't need the really fine I'm I don't need the really high-speed internal drive that the new high-speed internal drive that the new high-speed internal drive that the new generation provides. I'll just use an generation provides. I'll just use an generation provides. I'll just use an external drive and that's plenty fast external drive and that's plenty fast external drive and that's plenty fast already. And here we're seeing the already. And here we're seeing the already. And here we're seeing the numbers dip quite a bit on the Mac. numbers dip quite a bit on the Mac. numbers dip quite a bit on the Mac. 6,960 6,960 6,960 for read, 6,600 for right. The random for read, 6,600 for right. The random for read, 6,600 for right. The random numbers are actually about the same. But numbers are actually about the same. But numbers are actually about the same. But on the Razer, we are seeing quite a bit on the Razer, we are seeing quite a bit on the Razer, we are seeing quite a bit of a dip, especially on that right. of a dip, especially on that right. of a dip, especially on that right. 4,000 or so on read. I'm going to do 4,000 or so on read. I'm going to do 4,000 or so on read. I'm going to do that again because that seems really that again because that seems really that again because that seems really low. 1300 is what we got on right, which low. 1300 is what we got on right, which low. 1300 is what we got on right, which is just kind of pretty low. The read is just kind of pretty low. The read is just kind of pretty low. The read speed repeated the second time and right speed repeated the second time and right speed repeated the second time and right speed is about 1300. So, not great. speed is about 1300. So, not great. speed is about 1300. So, not great. [music] Now, some of you might have [music] Now, some of you might have [music] Now, some of you might have noticed that I had this plugged into the noticed that I had this plugged into the noticed that I had this plugged into the left side. By the way, these ports are left side. By the way, these ports are left side. By the way, these ports are not labeled. There's a USBC port here, not labeled. There's a USBC port here, not labeled. There's a USBC port here, and there's one on the right. They're and there's one on the right. They're and there's one on the right. They're not labeled. You have to know that the not labeled. You have to know that the not labeled. You have to know that the one on the right is the Thunderbolt 5 one on the right is the Thunderbolt 5 one on the right is the Thunderbolt 5 one. And sure enough, when I plugged the one. And sure enough, when I plugged the one. And sure enough, when I plugged the external drive into that, ran this external drive into that, ran this external drive into that, ran this again, I got a higher read score for again, I got a higher read score for again, I got a higher read score for sequential, but the right speed 1300.

sequential, but the right speed 1300. sequential, but the right speed 1300. Again, really better random score on Again, really better random score on Again, really better random score on just this bottom one, but overall random just this bottom one, but overall random just this bottom one, but overall random score is lower than the Mac. So, they score is lower than the Mac. So, they score is lower than the Mac. So, they still have some work to do as far as still have some work to do as far as still have some work to do as far as their Thunderbolt support. And this is their Thunderbolt support. And this is their Thunderbolt support. And this is an Intel box, so they need to get a move an Intel box, so they need to get a move an Intel box, so they need to get a move on that. Speaking of moving on, I'm on that. Speaking of moving on, I'm on that. Speaking of moving on, I'm going to move on going to move on going to move on to uh some multi-core tests. This is the to uh some multi-core tests. This is the to uh some multi-core tests. This is the Mattelroad algorithm implemented in Mattelroad algorithm implemented in Mattelroad algorithm implemented in Python and it's a pretty intense Python and it's a pretty intense Python and it's a pretty intense algorithm. It uses up all the available algorithm. It uses up all the available algorithm. It uses up all the available cores and just pegs them to the max. cores and just pegs them to the max. cores and just pegs them to the max. Now, I'm going to send the 1600 Now, I'm going to send the 1600 Now, I'm going to send the 1600 parameter and send the rest to dev null. parameter and send the rest to dev null. parameter and send the rest to dev null. We don't see output. I'm using the time We don't see output. I'm using the time We don't see output. I'm using the time command on the Mac and the measure command on the Mac and the measure command on the Mac and the measure command on Windows in PowerShell so that command on Windows in PowerShell so that command on Windows in PowerShell so that we get the timing of each one. And let's we get the timing of each one. And let's we get the timing of each one. And let's go. And now go. And now go. And now I total my thumbs. I usually cut this I total my thumbs. I usually cut this I total my thumbs. I usually cut this part out, but hey, it's done. Wow. Okay, part out, but hey, it's done. Wow. Okay, part out, but hey, it's done. Wow. Okay, it's fast on both of them. That was only it's fast on both of them. That was only it's fast on both of them. That was only 11.3 seconds on the Mac and 13.5 seconds 11.3 seconds on the Mac and 13.5 seconds 11.3 seconds on the Mac and 13.5 seconds on the Windows machine. Both really, on the Windows machine. Both really, on the Windows machine. Both really, really fast. By the way, that's what it really fast. By the way, that's what it really fast. By the way, that's what it looks like when running on the Windows looks like when running on the Windows looks like when running on the Windows machine. Every one of those cores was machine. Every one of those cores was machine. Every one of those cores was pegged to the max at that point in time.

pegged to the max at that point in time. pegged to the max at that point in time. And just so you can see what it looks And just so you can see what it looks And just so you can see what it looks like on the Mac here, we got 12 like on the Mac here, we got 12 like on the Mac here, we got 12 performance cores and six super cores. performance cores and six super cores. performance cores and six super cores. You can see the green soldiers marching. You can see the green soldiers marching. You can see the green soldiers marching. That's basically what it looks like That's basically what it looks like That's basically what it looks like right there. Very similar pattern. All right there. Very similar pattern. All right there. Very similar pattern. All the cores being used, but we got six the cores being used, but we got six the cores being used, but we got six less cores on the Mac than we do on the less cores on the Mac than we do on the less cores on the Mac than we do on the Windows machine. Now, let's do some Windows machine. Now, let's do some Windows machine. Now, let's do some compilations. Damen Edwards uh created compilations. Damen Edwards uh created compilations. Damen Edwards uh created this repository withnet compilations. this repository withnet compilations. this repository withnet compilations. Damian Edwards is an architect on the Damian Edwards is an architect on the Damian Edwards is an architect on the thennet team and it's called Devbench. thennet team and it's called Devbench. thennet team and it's called Devbench. Now this one has a couple of different Now this one has a couple of different Now this one has a couple of different options to run it. There's a hello world options to run it. There's a hello world options to run it. There's a hello world which we're going to start with and it's which we're going to start with and it's which we're going to start with and it's a small little net application build. a small little net application build. a small little net application build. Shouldn't take long at all and I Shouldn't take long at all and I Shouldn't take long at all and I suspected the Razer is going to win and suspected the Razer is going to win and suspected the Razer is going to win and uh as I was saying that it actually uh as I was saying that it actually uh as I was saying that it actually finished but the Mac won. finished but the Mac won. finished but the Mac won. Um, so what's nice about this benchmark, Um, so what's nice about this benchmark, Um, so what's nice about this benchmark, we're going to do it again, don't worry, we're going to do it again, don't worry, we're going to do it again, don't worry, is that it does a cold build, a warm is that it does a cold build, a warm is that it does a cold build, a warm build, and then an incremental. And on build, and then an incremental. And on build, and then an incremental. And on each one of those, the MacBook one. each one of those, the MacBook one. each one of those, the MacBook one. Okay, so hello world didn't work out for Okay, so hello world didn't work out for Okay, so hello world didn't work out for us on the Razer side. Let's try this us on the Razer side. Let's try this us on the Razer side. Let's try this next one, which is actually a large next one, which is actually a large next one, which is actually a large project that's 100,000 classes and project that's 100,000 classes and project that's 100,000 classes and namespaces. Boom. This is going to take namespaces. Boom. This is going to take namespaces. Boom. This is going to take a little bit longer. Not that long, but a little bit longer. Not that long, but a little bit longer. Not that long, but this represents a pretty large project.

this represents a pretty large project. this represents a pretty large project. And we're going to do an actual build of And we're going to do an actual build of And we're going to do an actual build of a real project next. All right, the cold a real project next. All right, the cold a real project next. All right, the cold build finished in 45 seconds here on the build finished in 45 seconds here on the build finished in 45 seconds here on the Mac. Warm build 8.9 seconds. And you can Mac. Warm build 8.9 seconds. And you can Mac. Warm build 8.9 seconds. And you can see how it's hitting the CPU cores. It's see how it's hitting the CPU cores. It's see how it's hitting the CPU cores. It's kind of all over the place. That's why kind of all over the place. That's why kind of all over the place. That's why this is a more realistic example of an this is a more realistic example of an this is a more realistic example of an actual build. And this is using the actual build. And this is using the actual build. And this is using the disc. Wo, not what I expected. 56 disc. Wo, not what I expected. 56 disc. Wo, not what I expected. 56 seconds for the Razer for the cold build seconds for the Razer for the cold build seconds for the Razer for the cold build and 15 seconds for the warm build. two and 15 seconds for the warm build. two and 15 seconds for the warm build. two times longer. All right, let's do one times longer. All right, let's do one times longer. All right, let's do one more. This is the real CMS project more. This is the real CMS project more. This is the real CMS project called Orchard. And by the way, this called Orchard. And by the way, this called Orchard. And by the way, this repository is available. You can clone repository is available. You can clone repository is available. You can clone this yourself and build it yourself if this yourself and build it yourself if this yourself and build it yourself if you want to try it out on your system. you want to try it out on your system. you want to try it out on your system. See how it goes for you. I'll link to it See how it goes for you. I'll link to it See how it goes for you. I'll link to it down below. Yeah. So, down below. Yeah. So, down below. Yeah. So, this uh this proves it. Cold build on this uh this proves it. Cold build on this uh this proves it. Cold build on the Razer took 42 seconds, a warm build the Razer took 42 seconds, a warm build the Razer took 42 seconds, a warm build 9 seconds, and an incremental 7.9. 9 seconds, and an incremental 7.9. 9 seconds, and an incremental 7.9. Everything is faster on the Mac. 21.8 8 Everything is faster on the Mac. 21.8 8 Everything is faster on the Mac. 21.8 8 seconds for the cold build. Two times seconds for the cold build. Two times seconds for the cold build. Two times faster. The warm belt 6 seconds and faster. The warm belt 6 seconds and faster. The warm belt 6 seconds and incremental 5.9. This was kind of like incremental 5.9. This was kind of like incremental 5.9. This was kind of like the test that I really thought that the the test that I really thought that the the test that I really thought that the Razer would win this time, but it did Razer would win this time, but it did Razer would win this time, but it did not happen. But there's still something not happen. But there's still something not happen. But there's still something that Razer has in its pocket that that Razer has in its pocket that that Razer has in its pocket that [music] might redeem it, and that's that [music] might redeem it, and that's that [music] might redeem it, and that's that RTX 5090. Now, I did mention we'll get RTX 5090. Now, I did mention we'll get RTX 5090. Now, I did mention we'll get back to Windows and Visual Studio. So back to Windows and Visual Studio. So back to Windows and Visual Studio. So here I have Visual Studio which is a here I have Visual Studio which is a here I have Visual Studio which is a Windows only program since Visual Studio Windows only program since Visual Studio Windows only program since Visual Studio for Mac was executed a little while ago.

for Mac was executed a little while ago. for Mac was executed a little while ago. Let's see how long it takes to create a Let's see how long it takes to create a Let's see how long it takes to create a new project. Last year if you remember new project. Last year if you remember new project. Last year if you remember if you saw that video it actually was if you saw that video it actually was if you saw that video it actually was faster to create a new project and to faster to create a new project and to faster to create a new project and to start it up in a virtual machine on the start it up in a virtual machine on the start it up in a virtual machine on the Mac instead of Windows. So let's see. Mac instead of Windows. So let's see. Mac instead of Windows. So let's see. I'm on the last step. I'm going to hit I'm on the last step. I'm going to hit I'm on the last step. I'm going to hit create and we'll see which one pops up create and we'll see which one pops up create and we'll see which one pops up first. first. first. [music] Come on. Come on. It's creating. Oh, I Come on. Come on. It's creating. Oh, I think that this one actually did it. think that this one actually did it. think that this one actually did it. Nvidia usage collection, which is don't Nvidia usage collection, which is don't Nvidia usage collection, which is don't know what that was all about, but it know what that was all about, but it know what that was all about, but it seemed to me like the Razer actually seemed to me like the Razer actually seemed to me like the Razer actually created the project faster. Not that created the project faster. Not that created the project faster. Not that much faster, but faster. Of course, this much faster, but faster. Of course, this much faster, but faster. Of course, this creates a Hello World Blazer creates a Hello World Blazer creates a Hello World Blazer application. And boom. And we should get application. And boom. And we should get application. And boom. And we should get some questions if they didn't fix that some questions if they didn't fix that some questions if they didn't fix that yet. Yep, there we go. Project yet. Yep, there we go. Project yet. Yep, there we go. Project configured to use SSL. Don't ask me configured to use SSL. Don't ask me configured to use SSL. Don't ask me again. It's going to ask me again. It's again. It's going to ask me again. It's again. It's going to ask me again. It's going to ask me again. Yes. going to ask me again. Yes. going to ask me again. Yes. Yes. Don't ask me again. Yes. Don't ask me again. Yes. Don't ask me again. >> [laughter] >> [laughter] >> [laughter] >> Notice a pattern. Boom. And one more >> Notice a pattern. Boom. And one more >> Notice a pattern. Boom. And one more time. Four times. Boom. Now we're time. Four times. Boom. Now we're time. Four times. Boom. Now we're waiting for that browser to pop up. And waiting for that browser to pop up. And waiting for that browser to pop up. And yes, the Razer actually wins this one by yes, the Razer actually wins this one by yes, the Razer actually wins this one by quite a bit this time around. Nice work.

quite a bit this time around. Nice work. quite a bit this time around. Nice work. That's how quickly it's supposed to pop That's how quickly it's supposed to pop That's how quickly it's supposed to pop up. I'd like it faster, of course, but up. I'd like it faster, of course, but up. I'd like it faster, of course, but you know. Now, let's do a subsequent you know. Now, let's do a subsequent you know. Now, let's do a subsequent build cuz it's already warm. And boom. build cuz it's already warm. And boom. build cuz it's already warm. And boom. [music] [music] [music] Now, they're about the same, but I think Now, they're about the same, but I think Now, they're about the same, but I think Razer is just a tad bit faster. Ladies Razer is just a tad bit faster. Ladies Razer is just a tad bit faster. Ladies and gentlemen, we're going to do some and gentlemen, we're going to do some and gentlemen, we're going to do some AI. I'm going to kick things off with a AI. I'm going to kick things off with a AI. I'm going to kick things off with a very simple test. We're going to do LM very simple test. We're going to do LM very simple test. We're going to do LM Studio, and this is Quen 34B, the Studio, and this is Quen 34B, the Studio, and this is Quen 34B, the multimodal version. And here I've got multimodal version. And here I've got multimodal version. And here I've got the MLX version. Here I've got the GGUF the MLX version. Here I've got the GGUF the MLX version. Here I've got the GGUF version, but they're both 4bit version, but they're both 4bit version, but they're both 4bit quantizations. I'm running MLX on the quantizations. I'm running MLX on the quantizations. I'm running MLX on the Mac because that's optimized for the Mac because that's optimized for the Mac because that's optimized for the Mac. Let's crank up that context length Mac. Let's crank up that context length Mac. Let's crank up that context length to 262,000. Load the model and try my to 262,000. Load the model and try my to 262,000. Load the model and try my first prompt. Hi. Boom. Hi. How are you? first prompt. Hi. Boom. Hi. How are you? first prompt. Hi. Boom. Hi. How are you? Oh my gosh. What happened? 25 tokens per Oh my gosh. What happened? 25 tokens per Oh my gosh. What happened? 25 tokens per second. There's something wrong here. second. There's something wrong here. second. There's something wrong here. It's 181 tokens per second on the Mac. It's 181 tokens per second on the Mac. It's 181 tokens per second on the Mac. This is a small model. It should be way This is a small model. It should be way This is a small model. It should be way faster here. Let's give LM Studio a kick faster here. Let's give LM Studio a kick faster here. Let's give LM Studio a kick in the butt and try this one more time.

in the butt and try this one more time. in the butt and try this one more time. 21.9 tokens per second. And this, I'm 21.9 tokens per second. And this, I'm 21.9 tokens per second. And this, I'm afraid, is the limitation of the amount afraid, is the limitation of the amount afraid, is the limitation of the amount of memory we have. If you take a look at of memory we have. If you take a look at of memory we have. If you take a look at the GPU here, you'll see that it is the GPU here, you'll see that it is the GPU here, you'll see that it is running on the GPU, but it's also using running on the GPU, but it's also using running on the GPU, but it's also using up all the available memory of the GPU up all the available memory of the GPU up all the available memory of the GPU right there. 23.5 out of 24. And some of right there. 23.5 out of 24. And some of right there. 23.5 out of 24. And some of that stuff is spilling over to the CPU. that stuff is spilling over to the CPU. that stuff is spilling over to the CPU. That's why it's taking a while. While we That's why it's taking a while. While we That's why it's taking a while. While we do have 24 GB of VRAM available here to do have 24 GB of VRAM available here to do have 24 GB of VRAM available here to use, the context size that I specified use, the context size that I specified use, the context size that I specified eats all of that up. The estimated eats all of that up. The estimated eats all of that up. The estimated memory usage, LM Studio is telling you memory usage, LM Studio is telling you memory usage, LM Studio is telling you right there. All of it. All of it. So, I right there. All of it. All of it. So, I right there. All of it. All of it. So, I need to lower the context length to need to lower the context length to need to lower the context length to something more usable. I don't know. something more usable. I don't know. something more usable. I don't know. Let's go with 68,000. Ah, Let's go with 68,000. Ah, Let's go with 68,000. Ah, now we're cooking. Hello. I know you're now we're cooking. Hello. I know you're now we're cooking. Hello. I know you're sick of my small prompts, but it proves sick of my small prompts, but it proves sick of my small prompts, but it proves a point. It shows you at least right a point. It shows you at least right a point. It shows you at least right away how fast things are going. So, away how fast things are going. So, away how fast things are going. So, context length definitely matters here. context length definitely matters here. context length definitely matters here. And we're reaching about 129 to 160 And we're reaching about 129 to 160 And we're reaching about 129 to 160 tokens per second. For some reason, it's tokens per second. For some reason, it's tokens per second. For some reason, it's not getting consistent results. Here's not getting consistent results. Here's not getting consistent results. Here's 168. Not bad. Here we're getting 181.

168. Not bad. Here we're getting 181. 168. Not bad. Here we're getting 181. And I specified the full context length And I specified the full context length And I specified the full context length of 262,000. And the reason this is able of 262,000. And the reason this is able of 262,000. And the reason this is able to do it without even blinking is to do it without even blinking is to do it without even blinking is because this Mac has 128 GB of memory, because this Mac has 128 GB of memory, because this Mac has 128 GB of memory, most of which it can share with the GPU. most of which it can share with the GPU. most of which it can share with the GPU. And I've made many videos about that. And I've made many videos about that. And I've made many videos about that. Unified memory is what it's called. It's Unified memory is what it's called. It's Unified memory is what it's called. It's where the CPU and the GPU draw from the where the CPU and the GPU draw from the where the CPU and the GPU draw from the same big pool of memory. Well, on this same big pool of memory. Well, on this same big pool of memory. Well, on this computer, it's a big pool. MacBooks also computer, it's a big pool. MacBooks also computer, it's a big pool. MacBooks also come in 16 gigabyte varieties where they come in 16 gigabyte varieties where they come in 16 gigabyte varieties where they don't have even as much as this machine. don't have even as much as this machine. don't have even as much as this machine. So that's MLX and LM Studio. Now I did So that's MLX and LM Studio. Now I did So that's MLX and LM Studio. Now I did also run slightly larger models and I also run slightly larger models and I also run slightly larger models and I ran the GGUF versions in Llama CPP which ran the GGUF versions in Llama CPP which ran the GGUF versions in Llama CPP which is probably the most popular way to run is probably the most popular way to run is probably the most popular way to run your models right now. I mean unless your models right now. I mean unless your models right now. I mean unless you're serving VLM and a lot of users. you're serving VLM and a lot of users. you're serving VLM and a lot of users. That's a topic for another day and not That's a topic for another day and not That's a topic for another day and not related to running local AI on your related to running local AI on your related to running local AI on your machines. You can see my other videos machines. You can see my other videos machines. You can see my other videos with bigger boxes for examples of that. with bigger boxes for examples of that. with bigger boxes for examples of that. For now, I can see that prompt For now, I can see that prompt For now, I can see that prompt processing, which is the initial phase processing, which is the initial phase processing, which is the initial phase of inference, is much faster on the of inference, is much faster on the of inference, is much faster on the Razer. When it comes to Gemma 412B, Quen Razer. When it comes to Gemma 412B, Quen Razer. When it comes to Gemma 412B, Quen 3.527B, 3.527B, 3.527B, which is a dense model, that's why the which is a dense model, that's why the which is a dense model, that's why the numbers are a little bit lower here. And numbers are a little bit lower here. And numbers are a little bit lower here. And Quen 3.635B, Quen 3.635B, Quen 3.635B, another mixture of experts. That one has another mixture of experts. That one has another mixture of experts. That one has 4,235 4,235 4,235 tokens per second for prompt processing tokens per second for prompt processing tokens per second for prompt processing versus the max 3,215.

versus the max 3,215. versus the max 3,215. And finally, token generation. Again, And finally, token generation. Again, And finally, token generation. Again, the Razer wins here, and it's not even the Razer wins here, and it's not even the Razer wins here, and it's not even close. Quen 3.69 6 169.1 tokens per close. Quen 3.69 6 169.1 tokens per close. Quen 3.69 6 169.1 tokens per second on the Razer, 95 tokens per second on the Razer, 95 tokens per second on the Razer, 95 tokens per second on the Mac. And you can see Gemma second on the Mac. And you can see Gemma second on the Mac. And you can see Gemma 4 and Quinn 3.5 here as well. Last year 4 and Quinn 3.5 here as well. Last year 4 and Quinn 3.5 here as well. Last year I said this machine couldn't justify its I said this machine couldn't justify its I said this machine couldn't justify its price. This year on this test, it price. This year on this test, it price. This year on this test, it absolutely does. But the Mac has another absolutely does. But the Mac has another absolutely does. But the Mac has another trick up its sleeve. Here I'm spinning trick up its sleeve. Here I'm spinning trick up its sleeve. Here I'm spinning up Llama CLI and I'm giving it a prompt up Llama CLI and I'm giving it a prompt up Llama CLI and I'm giving it a prompt on the terminal. But notice the model on the terminal. But notice the model on the terminal. But notice the model I'm using. Quen 3.5 122 billion I'm using. Quen 3.5 122 billion I'm using. Quen 3.5 122 billion parameter model and it's printing out parameter model and it's printing out parameter model and it's printing out pretty fast. We're getting prompt pretty fast. We're getting prompt pretty fast. We're getting prompt processing of 139 tokens per second processing of 139 tokens per second processing of 139 tokens per second generation at 47 tokens per second right generation at 47 tokens per second right generation at 47 tokens per second right now. 94 GB of memory is being used on now. 94 GB of memory is being used on now. 94 GB of memory is being used on this machine. And there's that GPU, that this machine. And there's that GPU, that this machine. And there's that GPU, that blue line right there. That's the GPU blue line right there. That's the GPU blue line right there. That's the GPU usage during the generation. There's no usage during the generation. There's no usage during the generation. There's no way that any laptop can touch that kind way that any laptop can touch that kind way that any laptop can touch that kind of model unless you have a stricks Halo of model unless you have a stricks Halo of model unless you have a stricks Halo machine with 128 gigs of memory. or if machine with 128 gigs of memory. or if machine with 128 gigs of memory. or if you go to the desktop and you have an you go to the desktop and you have an you go to the desktop and you have an RTX Pro 6000 GPU, but a 5090, even a RTX Pro 6000 GPU, but a 5090, even a RTX Pro 6000 GPU, but a 5090, even a desktop 5090 with 32 gigs of memory, desktop 5090 with 32 gigs of memory, desktop 5090 with 32 gigs of memory, will not be able to run a model that will not be able to run a model that will not be able to run a model that large. So, overall, the Razer is a large. So, overall, the Razer is a large. So, overall, the Razer is a pretty impressive machine this year, you pretty impressive machine this year, you pretty impressive machine this year, you win some, you lose some, depends on what win some, you lose some, depends on what win some, you lose some, depends on what you're working on. If the things that you're working on. If the things that you're working on. If the things that you do on your laptop, include gaming you do on your laptop, include gaming you do on your laptop, include gaming and AI work, then this is definitely a and AI work, then this is definitely a and AI work, then this is definitely a good laptop to get, especially because good laptop to get, especially because good laptop to get, especially because this year it's wellb built, it's quiet, this year it's wellb built, it's quiet, this year it's wellb built, it's quiet, very quiet, it's got a beautiful screen.

very quiet, it's got a beautiful screen. very quiet, it's got a beautiful screen. There are, of course, things I dinged it There are, of course, things I dinged it There are, of course, things I dinged it for, but here's a thing to watch out for, but here's a thing to watch out for, but here's a thing to watch out for. If you're shopping, make sure that for. If you're shopping, make sure that for. If you're shopping, make sure that you're buying the 2026 model with the you're buying the 2026 model with the you're buying the 2026 model with the Intel Core Ultra 9 290HX, Intel Core Ultra 9 290HX, Intel Core Ultra 9 290HX, not the 2025 edition, which is going to not the 2025 edition, which is going to not the 2025 edition, which is going to disappoint you. I think you're curious disappoint you. I think you're curious disappoint you. I think you're curious about how the M4 Max does against the about how the M4 Max does against the about how the M4 Max does against the 2025 version, watch the video here. 2025 version, watch the video here. 2025 version, watch the video here. Otherwise, [music] thanks for watching Otherwise, [music] thanks for watching Otherwise, [music] thanks for watching and I'll see you next time.