Xiaomi MiMo V2.5 Pro a prova: codi 3D brillant, però una experiència desesperant

After all Okay, nope. All right. Well, it started off nice. Yeah, that's that's 100% a security camera. I don't I've never seen that level of detail in this result. Today, we're going to be looking at ShowMe Memo V2.5 Pro, which is the newest open-source model. Now, you may be thinking, "Vijan, this model's been out for like a few days now." Yes, however, they did just recently release the weights for this on Hugging Face. So, now, if we click this Hugging Face hyperlink, which we don't need to do cuz I do just have the tab open already, we will see that this model in its entirety is now hosted open-source with an MIT license for anyone interested in actually running this themselves. Now, the caveat on that is most people don't have the hardware to actually do so, but although we're not going to be testing it in this video, they did also release the V2.5 version. So, the difference here is this is a V2.5 Pro. This is V2.5, which is about a little more than a third smaller. And additionally to that, this is a native Omni model, so it is very likely that I will do a dedicated video on V2.5 as well. Though, for today, we're going to be solely focused on V2.5 Pro, as it does have the potential based off of the things I'm seeing to be the most performant open-source model currently existing, which is quite exciting. So, before we get into it, please do feel free to subscribe, as I do want that 100K plaque, and then I'll stop asking. No, I probably won't. I'll probably ask for like whatever the next plaque is that comes after that. But, let's begin by just taking a quick peek at the announcement blog post here, where of course we start with a benchmark. Not a benchmark JPEG, because this does actually have tooltips over the specific model that is showing benchmarked against here. But, this is apparently stacking up very favorably against closed-source state-of-the-art models. Now, we should note that Opus 4.6 and GPT-54 are now not the most current up-to-date variations of those specific models. However, at the time these benchmarks were done, they likely probably were. And we see right here, this does stack up pretty well. Now, I always like to just test the things and see how I feel based on that versus just judging based on benchmarks. Though, I will say I did test V2 Pro when it came out and it had one of the more impressive ship combat simulator results I'd seen where the water effects it drew in the ship models and just overall the quality of the results that we received in that test were extremely impressive. So, I'm really excited to test this new version of this model. In terms of some more tech specs stuff, we see right here that not only was the model itself released, but the base model for this was also released, which is pretty exciting because one could perhaps use this to make a baby Mythos style model. Now, I'm what? I'm not kidding cuz you could, but it's just, you know, it's cool to see base models there as well. We have some more additional benchmarks down here, but for now let's swap over to the Hugging Face model card for this specific Pro variant and we'll take a peek at it. So, we see this is a total size of just over 1 trillion total parameters. It is a mixture of experts model with 42 billion active. It has hybrid attention architecture and MTP. So, hypothetically it will be able to run a bit faster being that it has that

MTP, which speeds up the token generation speed to put it simply as opposed to a model that did not have that. Additionally to that, we see some more technical specifications right here. If we scroll on down, we see the context length is 1 million, which is quite a healthy context length. Now again, I don't specifically have tests to really stretch the context length and push it very far. It is something I'd like to work on, but it's just interesting to see models with that context length being more common than they used to be. So, in terms of pricing, if we assume at the maximum context length for the pricing, the input is $2 per million tokens and output is $6 per million output tokens. So, not a bad price and especially if this does really stack up as shown in some of the benchmark graphics that we see right here. This could potentially be a fairly cheaper (in parentheses) option comparatively to a closed-sourced expensive model. And we can see right here, I had to search for this for a little, but I did want to be able to find it. The maximum output length is 128k, so a healthy length. Now, our next step really is to hop into the Mimo Studio right here. I do have this hooked up in open code just through OpenRouter, so we will likely be doing some of the tests in that. But for a model like this, I always do mostly enjoy testing it just in the web chat interface because it provides the quickest access to like a variety of different tests to get a general feel for the model's capability across a diverse set of tasks. So, with that, we shall begin with our tried-and-true Browser OS Test V2.5. This is the version of the Browser OS Test where it has to create two of the five apps being functional 3D games, with one of them being a GTA clone and the other up to it, the ability to change wallpaper, the special feature that it decides on and implements. I might title this this model's ins- is insane, but like the the dictionary definition of the word. Just based off of this this thinking chain. Oh, wow. Okay, so after 663.4 seconds of thinking, it has begun to write the script. That's 10 minutes. Right? 60 seconds is a minute. 10 of those would be 600 seconds. Yeah, that's 11 minutes and 3 seconds. All right, well, this better be a good result. All right, so here is our Mimo V2 Pro Browser OS after 11 minutes of thinking. Okay. I can't say that I've ever seen anything quite like this because I have absolutely no idea what this is supposed to be. I will say that the flickering effect of it could perhaps be a potential lawsuit were this to be an actual piece of software that was delivered to folks just from like a medical standpoint, but it is definitely creative and I've never seen something like this. So, to begin, let's of course check if there's a right click. Okay, there is. Ah, toggle nebula. Good. I think maybe just for now we're going to leave that off for the rest of this video, but we can see that there is a

clock in the bottom right, which is showing the correct time in my locale, little after 4:00 a.m. Now, okay, let's just begin by going through first our start menu. Very good. Everything looks good. Everything looks nice and clean. I have no complaints here. Some models put in like a search feature here where you can search for an app, but that's not specifically listed as something it needs to have, so it's just something to notice. Now I'm curious though like Okay, well, I would This is inevitably going to be the special feature, I would imagine. Next wallpaper. Can we change wallpaper with the nebula on? Okay, I don't think we can. Okay, but we can change it from the right click there. It's actually It's not bad. All right, so I normally just start from like the first app and then move sequentially, but I'm always curious about the GTA results, so Okay, so our GTA app is not working. We're getting a ton of errors. Let me just toggle nebula off and we'll at least try the terminal first. I am going to get the GTA result to be fixed, so no worries about that because I want to see it. Let's just do calc. Okay, 4 + 4 8. Very good. We do have a working calculator. Let's do neofetch. Let's check out our system stats. Okay, that looks good. Omni OS. Okay, OMMS. We have 9.5 megabytes of memory, so perhaps a more resource constrained system, which could be the reason that the neon city app is not working, but we can do um This is a good looking terminal, I will say though. I do like the yellow and green color theme and whenever they put in some sort of like stylistic thing like this, I do like that. All right, good. And it even shows EDT. That's pretty cool. That is the correct time here. Can I minimize this to the taskbar and then reopen it from there? Good. That is nice to see. Next up, we have files. Documents. I like that. So, it has text file and we can actually open this. Can we open multiple at the same time? Okay, we can. Remember to try the nebula desktop mode in settings. Oh, believe me, I've tried that and I opened with it. Can we save this? Would save it to our system. I wonder if it will actually save this in the little fake file system that we have here. So, let's do save. And then we'll close that. So, if we open the notes. txt. Hey, that's cool. All right, I'll take that. It did properly save it. We have a pretty like deep file system right here. The way it goes all the way back. We have our temp files. We have home, which is where we have most of our things. We have Okay, so nothing there in pictures and nothing there in desktop. But overall, not a bad little emulated file system. And I did like that we saved the note and it actually was able to be opened back up from its location in the file system. Next up, calculator. Let's do 95 * 9. 855. All right, divided by 9. 95. Not bad. Next up, we have our settings application where we can cycle through

the wallpapers here. And they were nice. Now, it's funny like it's almost like all of a sudden every single model decided to just do gradients now for the multiple wallpapers instead of using actual photos from Unsplash or something. It's just interesting to see that like odd progression. Nebula desktop interactive particles. And we can of course toggle that on. Definitely a special feature. I'll leave it at that. We did look at our text editor already. Everything was nice there. Unfortunately, neon city did not work. Let's see if void assault does. It does. Well, you know what? I like this and this gives me some pretty bullish hope for actually getting GTA to work assuming we can. It's pseudo three actually no, that's not pseudo 3D. I'd say that's 3D. We can't move up and down. We can only move side to side. Maybe it's pseudo 3D. No, cuz the asteroids are visible. Okay, this isn't bad. Now unfortunately the big issue is that we have a bunch of errors showing up in the developer tools console right here. If we open that, we see there are many. I am sort of wondering if I should just put that in its own little folder and then have it go through open code and actually try to fix this. I think that's what I'm going to do. All right, so I had gone into open code and spent a little bit of time basically having to give this a few different tries to actually fix the GTA issue result. I found that it responded significantly better to just sending it the specific developer tool console error as opposed to a plain English description of the issue basically saying it's a black screen fix it. So now let's hopefully see. I will switch out of the nebula mode. Hopefully our G TA result Oh my. Um well, it fixed it. The issue is just it's so dark. We can't see anything. But hey, there's a person. This looks like it may potentially be rather impressive. Unfortunately, it's too dark to tell. Though that should likely be a fairly simple fix in the scheme of things. So I think I spelled environment wrong. That's okay. We'll see what we get here. All right, let's see if the changes it made to the brightness of the game do allow it to be more visible. Okay, yes. Not to an extreme degree, but somewhat. Okay, so we can walk around the city and this is definitely a neon city theme. We even have a mini-map which we didn't ask for. It's just something it added in on its own. F would be to enter and exit a car. Now, I don't know that I've been able to see a car so far. All right. So, I can hypothetically steal this car by pressing F. And we did. It just changes the camera view, so we're Okay, that's cool. It actually brought the car model with us. So, you know, this is Oh, there. Okay. This is an interesting take. I will say I think I was expecting a little better just in terms of the fact that it didn't work on the first, second, or third tries. I had to fix it in open code. Though, the neon vibe is actually pretty cool. The mini map is neat. And the car

models are actually kind of endearing in a way. They have headlights. They have nice car-like shapes. Do they have tail lights? Okay, they don't have tail lights. That's going to be our new right click. It's do they have tail lights? No. Um, I do like the neon in the buildings and things. So, it's an interesting result. And something I'll say about the browser OS here is it's very unique in terms of some of the graphical decisions that it chose to implement here. A very, very niche result. So, next up we're going to be running the beautiful static subway scene test. It's always fun to see what sort of models they make. And then, in the follow-up test to this, we'll have it turn whatever it creates into an FPS just using that map. All right. So, unfortunately, and quite frustratingly, right after 735 seconds of thinking, it started generating the script and then said I can't answer this question for now. Let's talk about something else. So, I've kind of questioned its decision. And we'll see what it does. The user is frustrated and angry. Something went wrong on my end. Uh I wonder if it nerfs the context window if it triggers this. I'm just going to just going to do it again. Don't overthink. Just just to put that in there. We'll see what happens. In the meantime, though, because uh I am extremely enraged by what has previously happened, I am going to now through What the heck? Okay. So, by adding don't overthink to the prompt, it went from 730 seconds of thought process to 3.1. That was one of perhaps the most noticeable changes I've ever seen in a model's behavior through the addition of two words. So, this should soon actually just render this in an artifact window. I'm going to allow it to do that just to make sure this actually does work because if it does, okay, this hypothetically looks incredibly impressive. Let's check our brightness slider. Spot-on and one of the better subway car models I've seen, I will say. Now, the thing that I'm going to do, we're not going to look at it entirely through here. I will download it and we'll take a peek at it in Chrome. However, I'm going to now send a follow-up prompt saying, "Awesome, now turn this into a FPS." So, I'm sending that follow-up. Oh, I didn't specifically denote the don't overthink, but that's okay. We have a little bit of time here. Let's just Let's take a look at the subway station now natively. So, brightness initially, obviously, is too low, but something that models sometimes will fail to do is there's no actual ability to get to the brightness slider because as soon as you get into the scene, you can't actually select that and they'll have like a lock thing over here. So, there will be some UI element here and it will inhibit you from selecting the brightness slider. TLDR, the brightness slider is accessible. That's not always the case. Now, something I would Is that a security camera? It is. And it's Is it? I think it's like a decent one, too.

Yeah, that's that's 100% a security camera. I don't I've never seen that level of detail in this result, I guess, could be said. We do have some signs. Okay, we have the gates right there. We have a stairway to nowhere. And then we have a what appears to be a vending or a ticket machine or something of the sort. Really, you know what? This is nice. It's like lightweight, too, I guess could be said. And what I mean by that is the fans on the laptop are not spinning as rapidly as they normally are when looking at the results generated for this specific prompt from other models. We have benches. We have this which I don't specifically know what it is. The subway model's actually pretty good. The yellow do not cross line. Ceiling looks all right as do the lights. I am pleased with this result and I've never seen security cameras. We have flight conduit running on the ceiling, which is good to see. So, like things for electrical tubing and piping. And this is overall it's clean. I like it. All right, so we did get our subway FPS created just based off of the subway result. So, I'll be very interested to see how this looks. Let's go ahead. Subway dead. Okay, let's Oh, wow. I mean Okay. This is good. I wonder if it put any sound in. I do have a speaker here. This is actually it's more eerie. It did put sound in. Wait, let me turn my speaker up. Okay, good. So, check this out. Look at the score in the top right. Wave two, they're coming. It's weird. It took the stairs away. All right, this is good. Look at the weapon recoil. The ammunition tracers are there and they look really good. The only thing I've never seen except for the Claude Opus 4.7 result was the ammunition actually was the ammo actually like showing if you just hit an artifact of the environment. Uh-oh, I'm out. How do I reload? Uh-oh. Okay, you know what? That was darn good. That was really a nice transposition of the original original subway station result here, and then it took that and turned it into this, which is like This is quite good. All right, I'm happy. I'm now going to give this the task to create an accurate 3JS model of Jerry's apartment from Seinfeld. I don't know if it will be able to do this or not. There is no photo reference given, so just keep that in mind here. Now, unfortunately, I seem to be having some issues because I am also currently running the C++ skateboard simulation test. The reason that I did not specifically showcase actually beginning this or doing anything with it in open code is because I keep getting provider issues, and I'm using this through open router. So, it's consistently basically shown an error

that would be indicative of a network issue, which there isn't. So, it may just not be like good luck right now using it through there. So, I'm going to do my best to try to get that as well, but for now we're going to focus on the Seinfeld's apartment test. All right, so we did get our Jerry's apartment result. Now, it did show up initially in the preview pane, but fortunately it's outside the apartment, so no there are no spoilers here. Let's take a peek at this. This thought for again probably like 700 something seconds. So, this is a very deeply thoughtful model, I guess could be said. All right, I don't know why I'm like like scared to go in. All right, let's do it. Through the door. You know what? This doesn't look like Jerry's apartment, but this is giving me like old person's Florida home vibes, and I kind of don't mind it. So, okay. We have a very very interesting chair right here, as well as another interesting chair placed in the table. We do have the couch. We have some nice art on the wall there. I just need to kind of figure out how to navigate around this scene. Okay, we have the kitchen. Everything is very clean and sterile. You know, this reminds me of like an IKEA like demo space if you've seen that. We have some cereal boxes there. We have an oven. We have a refrigerator. We have a Superman poster on the wall. We even do have the bicycle, although it is red. And it is a little difficult to actually see it. It's over there. The lights look good. I would say if this was not specifically designed to emulate an apartment, which it looks nothing like, I would have to say this is definitely showing promise as it's very clean and the assets are drawn in a nice manner, I guess could be said. This is something where you could definitely use this for 3D 3JS scenes like this and it would be able to produce some good-looking assets. Not 100% sure what's going on with this obscure pink chair here, but aside from that, everything else in this scene is actually cool. They're even like the handles on the cabinets. There's some material detail like the chrome for this lamp base or for that light fixture base. We have a door there. Let's see, did it attempt to create any other like subspaces in here or not a subspace, but like the bathroom or the bedroom or something? Okay, I don't believe it did. There is one door that was on that side. So, if we go in there, No. So, okay. An interesting result, not necessarily what we wanted in terms of the layout, but it did do a good job in showcasing its ability to do 3D scenes like this. So, for our next prompt, we're going to be giving the classic tried and true flight combat simulator game test where it needs to build three selectable plane models that we can choose from and play and then we engage in some flight combat. It should be 3D and have some good-looking graphical attributes to it. And I had worried that it was going to give me this refusal here. It seems to be very, very sensitive in triggering this, which is very frustrating, but

I'm just going to send it the same combat simulator result again, but I'm going to add don't overthink. And again, we just we're getting this refusal, which is just it's like unusable. So, I'd run the flight combat simulator test with V2 Pro. We can see right here, this is from when I was filming the video for the V2 Pro, and it did produce a result. Unfortunately, it seems like V2.5 Pro is just far more inclined to cause some form of issues. So, I'm actually going to tell it to improve this result massively because if we can't get it from a fresh chat window here, at least we can perhaps have 2.5 Pro iterate on something that 2 Pro did actually create, I guess. And I am including don't overthink because if it is going to trigger that refusal, I'd rather just have it do it off the bat than after thinking for like 10 minutes and then doing it. Oh, very good. Okay, so that was a very, very short amount of time to think. Now, I do have to admit I don't actually 100% remember what exactly it produced here for this result. So, I suppose it would be good to actually take a peek at this real quick just so we have some reference point for what we're having this improve. Oh. Well, I would say then we definitely have a good blank canvas for this to improve on. All right, well, I guess maybe this wasn't even in that V2 Pro video, which now is very possible, I would say. So, good. We'll have 2.5 hopefully really uh fix this up for us. All right, so we got some key improvements over the previous version. Now, there was no previous version because, quite frankly, it didn't work. So, I'm just going to quickly test this here. Okay. It's troubled, but um Oh. All right. All right, we have something to iterate on and that's a good starting point. This is absolute garbage. The user is extremely frustrated. The planes look like garbage. Okay, so I didn't say that. This is the core issue with my previous attempts. This is the model filling these things in itself. The planes look like garbage. They're just boxes stacked together. Voxel style doesn't mean random boxes. Okay. The flight model feels terrible. The combat is shallow. The presentation is poor. The terrain is boring. Okay. Wow, this thing really went hard on itself. I didn't specifically pick out any one of those items. I just said this is garbage. So it picked those out on its own. I did not tell it here don't overthink. So we may have to wait a little while for this result, but that's fine if we do get something improved. It seems like the thinking may have glitched out because it's still thinking, but the thinking process showed that it had finished here. So I'm not 100% sure what's going to happen here. I'm just going to leave this tab open and fingers crossed we do

actually get a iteration on this result. Oh, no. This model is so frustrating, honestly. I started filming this video hours ago and it's just prompt tokens, completion tokens, total tokens. It's like after waiting 10-15 minutes for it to think it actually responds. Now again, I have no justification to complain because it's free to use right here from this web chat interface, but it really is like borderline unusable and this is so frustrating because this is a polar opposite experience that I've had with this version of the model compared to the V2 Pro. And unfortunately, I've tried multiple times to get the self-contained C++ skateboarding game to result um to complete just through open router through open code, and every time it just basically stops or it just I'm not having much luck. All right. So, again, for some reason, we're just getting this refusal, which is basically rendered this unusable at this point. So, I don't 100% know what to do. I'm trying a 3D printer simulation test through the web chat interface. I have said don't overthink, just write the script, and we'll see if we actually get something here that is usable. Fingers crossed we do. So, consistently, I'm not able to get any more results from this. Um I don't want to put out a video that's like only shown a few prompts, but like at the same time, should I just delete the footage? I want to showcase this model cuz of course it it has huge potential. It's just so frustrating. Okay, now it's generating a result here apparently for the 3D printer simulation. We'll see what happens. All right. So, we did actually get a fully interactive 3D printer simulation result. Let's download this and take a peek at it. All right. So far, I find this to be acceptable. The only weird issue is like for some reason, just like this, it makes me uncomfortable, and I can't explain why. I just find this specific thing to be discomfort causing. Regardless, though, we have we have a nice model. There's no stupid fog, which sometimes does happen, and let's just go print and see how it gets. Look at that. Look, the spool is spooling. The spool is spinning. Very interesting. Now, okay, we're going to notice there's definitely some miscues here, but I did say do not overthink, and we also have to remember that this result was generated following a refusal after minutes of thinking. So, we could assume this perhaps may not be the optimal condition for this to showcase its full power, but I would say even still this result has significant levels of promise and I'm happy to report that this pool spinning is not something that is commonly seen at all. I've seen that a very very little amount of times before. Okay, let's just Oh wow. Okay, so this printer may need some servicing but it did do something. We have some

attempted infill and things of the sort. Let's reset uh Okay, let's reload the page and we'll just start a circle. Good. Fantastic. See and and imagine like the nozzle was just down there. Ooh, that's that's odd infill. It's like spider web style. I'll take it. After all Okay, nope. All right. Well, it started off nice and then it turned into like the Leaning Tower of Pisa. It seems like if we speed it up it just freaks out. It's like a real printer in that regard actually. So this could arguably be a more realistic take where the fast if it just goes too fast it it freaks. You know what? This is artistic. A result. So I'm sending the drum kit test here just through open router and I have also added don't overthink as that seems to just make things happen a bit faster and especially in a test like this there's no need for it to really be thinking too too long. Now let's see. Okay, so my drum kit simulation test is just I don't really see this. This is it's not going to it's just going to freeze. Okay, final code. Okay, final code. But wait, one more thing. Okay, final code. So through open router after probably like 15 or so minutes of thinking we did actually receive a drum kit simulation result which I'm very pleased to say. Okay. Uh now here's the thing. The prompt does specifically say it can be two and a half D photo realistic which this is. So I'm not going to be able to really knock it for its graphical decision here. Really, we're going to be focused on the sounds and if they work well. Let's see if our key map can be enabled. Okay, this looks nice. I'm happy with this and we also see them. So, let's just try. Good, it actually has different sounds. So, S is open hi-hat and D is closed hi-hat. All right. All right, not bad. Let's try. We did get a Phil Collins test, which Let's check out our auto play feature starting with rock. And it has a little indicator down there. Let's see if we slow it. Good. And the BPMs go by like five at a time, which is good. Next up, jazz. Funk. Oh, yeah, that's funky. Can I play along with it? Yep, I can. That's me on the ride. And then finally, electronic. Very, very electronic.

And look at the hit effect on the hi-hat. We can see it now. Let's Let's hop in here. All right. Overall, aside from the just unexpectedly photorealistic 2D assets, this is a completely competent result. Here's what I'm going to do. Let's send in a new prompt. And for that, we're going to go to ultimateplay.com, which is my own website, where you can upload and play and upload AI-generated games. And for this, we're going to find my low-poly GP game that was created with GPT-55 Pro, and I'm going to give it the specific prompt that gave us this end result here, which really was like just it was madness. So, I really do hope that this will properly do this without much issue. Oh, that's embarrassing. I had a spell check pasted into the prompt. I guess that was the specific prompt, so I don't want it to think for like 15 minutes and then hit that I'm sorry, I can't talk about this now. This is a massive request. Yeah, that's fair. Let me guess. Yep, yep. Hello, I can't answer this question for now. Let's talk about something else. You know what I'm going to do is I'm going to just copy-paste this right back to you because this worked before. I'm also going to just put it in Openrouter. This is a massive request. Good, it did say like the same thing that it started saying there when it was in its thinking process. So, inevitably, perhaps one of these will produce a proper result. I've received a completed result for the low-poly Moto GP game from Openrouter, and I'm extremely excited to see what we get here cuz this is a difficult task. Okay, let's just go with Fireblade. Interesting, Marco Rossi, Italy. Jake Thompson, USA. Yuki Tanaka, Japan. That's awesome. Like genuinely All right. So, inevitably we have a bit of an issue. Now, I noticed that this was actually good through Open Code. We did manage to get it working if we already had a script that was giving us an error, and we had a simple error to basically copy-paste into it. So, I am going to attempt to have this fixed this through Open Code, and I'm going to leave it in build mode, and we'll see what it does. So, hypothetically, it has been able to fix this through open code, and we should be able to Fantastic. Fantastic. You know what? Aw, okay. Now, the prompt says, okay, let's address the elephant in the room. The go thing just never went off the screen, which is infuriating, but the prompt does denote, I'm going to turn the speaker down a little, that the bike should actually perform a slight wheelie and then return to the ground.

And we see it's actually doing that here. Now, I wonder, is this bike model backwards? I think it may be. But it's it's weird cuz it's doing a wheelie in the front and then returning to the ground. The NPC bikes are actually doing fine. This is in some ways arguably better than the GPT 55 Pro result because there are no buildings on the track, which that one kept doing. But this is definitely more cartoon-like. But something I'm going to notice that I'm actually kind of impressed with, oh, is that I think these bike models are backwards. The sound is very it just is the same sound looped over, but as the gears go up, it like you can hear it's modified the sound to actually emulate how it would sound in those different gears. I will say I do like this little bottom right speedometer thing, which I'll have to move my head to actually be able to show. This is for a first try, actually I'm cool with this. So, I think this is definitely 100% worthy of getting some additional follow-ups here from within open code, assuming we have good luck with open router not glitching out or the error that I was getting earlier. So, I'd given it some feedback and we'll see if it's actually able to properly iterate on this result. I wouldn't 100% expect this to because I've had pretty poor luck in just using this so far through open code with open router just with connection issues and other things. But, in the meantime, we'll go back and actually see our update to the Moto GP game has actually concluded in open router, so that's pretty exciting. And basically, if we refresh this now, we should get a overhauled result. Okay, so maybe the overhaul was contained to the actual game itself. And it was. Good. It's better. I told it there was no visible track and basically just changed this from grass to sand on the ground. I guess that's okay. Let's make sure, do we have the opponents? We do. And the bike is actually in the correct orientation now. Though, I have to note, I do believe the particle effects that it had in the first Oh. I thought they had been removed, but they're still there. We have an exhaust there. We do have the wheelies. Good. And we have the first-person view, which is semi-first person, but still, that's a better angle, I guess, to actually be able to see. The sound is Let's get a look at one of the other bike models here. All right. Not bad. So, I'm asking it just for a beautiful front end for a website for the company called Slap Fashion. The design should be high-end New York City {slash} Tokyo style with modern elements like cards the user can interact with. I also mentioned it needs to have some fluid ripple effects in the hero section and a modern high-end design design firm style. Let's take a look at our Slap Fashion high-end website. Okay. Now, this is weird because

I would have expected like colors, not white and gold. You know what? Let's take a look at it and we'll see. Okay, I do have some That's a nice effect, I'm going to say, on that explore collection. As a matter of fact, the circle around the cursor actually expands and gets a little transparent when we get to that. That's a pretty cool effect. Okay, avant-garde streetwear. Whatever that means. Ready to wear. Limited edition. Current collections. Okay, now unfortunately, this is not loading in with any stock photos, which is really going to let this down, but I'll say the effects over the cards look good. This is giving me large throwbacks to some specific result that was generated with V2 Pro when doing that test video. So, fashion is not about clothes, it's about the space between who you are and who you become. Slap. That looks good. Established 2019. We're in 47 countries, we've crafted 12,000 pieces, and we've been active for 6 years. Curated selection. Okay, 1,280. So, Slap Fashion is more of a high-end middle price for fashion. We don't follow trends. We create tension between structure and fluidity, between silence and statement. The lookbook. Not bad. Join the inner circle. And that's a nicely done footer. Oh, I'm going to give it some feedback here. So, I've given it some feedback. Basically, I wanted to read you this, and additionally, if it's not able to actually put images in there as placeholder. I'm just going to send it again. We should now have a good-looking and properly improved Slap Fashion website result. Okay, significantly more in line with what I'd expected to see. Now, I did tell it to create SVGs for specific clothing items, and it really did, and that's actually I mean, this one is terrifying. It looks like a ghost, but this actually did a pretty decent job. The only thing I'm really going to knock this for now is there's so much empty space in this header that it's a little weird. If it had started kind of like right there, it would definitely look better. All right, let's see. Wear the future. This is a big, big improvement. Waxed utility pant. Look at that. I mean, the job it did in making up for not having image assets is very, very good. Slap. Okay, maybe not with this specific example, but it's very interesting because it kept the same kind of style of the site. It just reskinned the colors, which is acceptable. I mean, this definitely shows some promise, and I will say the winner here is definitely what it did in the absence of actual asset photos for creating clothing-like graphics. That was definitely exceeds expectations, specifically with the waxed utility pant. So, actually kind of cool. All right, so overall, our experience here, I would do have to say was just really let down at least by the AI Studio chat interface that I was using this with. It's almost, I will dare say, just unusable because it will think for 10,

20 minutes, and then right as it starts to generate a script, it will say, "I'm sorry, I can't talk about that now." For no reason, and it's just a huge waste of time to use it through there. I'm going to say completely honestly. Through Open Router, it worked a little better. The main thing I will pick up on with this model is that its thinking process is, um, as I watch it, I almost wonder like, "Is it it just about to go into a loop, or is it actually going to be able to break out of this?" Because continuously I would see things like, "Okay, time to write the code. But wait, also I need to do this. Okay, time to write the code." For all I know, that's like what GPT 5.5 Pro is doing, and they just hide that from us. So, I can't entirely discount that, and it does, when given the opportunity to not get messed up by the web chat interface, actually generate scripts. I just have to say the model was a bit frustrating to use because the user experience from the current providers I have available to be able to use it was a letdown. The model in and of itself seems totally competent and capable and I will say the few times I used it in an agentic coding thing properly through open code where I basically gave it some developer tool errors for certain things that were breaking sites. It fixed them very very quickly and seemed really good at actually fixing code based off of specific error messages that were causing the code to break. That was something that was exciting. Our browser OS had the most unique special feature of this nebula UI that I think I've ever seen and the GTA clone unfortunately did not work on first or second try. We had open code ultimately fix it and then brighten it up and it was actually pretty unique. The neon take on this with the NPCs, we could get in the cars and it did have somewhat realistic car models. It was endearing and I was actually kind of pleased with that. Like if we go here and then now we're actually in the car. Like that's actually kind of cool. They had headlights and stuff. So, a cool result. Next up and these are probably going to be a little out of order. So, let me try to figure out Oh, our subway station. This was fantastic. I was very happy with this. It had a very realistic subway model. Well, realistic in the style of what this is. It had a good brightness selection slider. Everything here looked good. It had security cameras which I've never in my life seen in a result like this. So, he really went to a level of depth there that was cool. Then finally our subway FPS game here. This was awesome. This really was absolutely awesome. This model even to just do like if if you're into this stuff like low poly game dev and stuff, it's fantastic. Look at the the recoil, the ammo tracers and things like that. I very very much was pleased with this result. I really enjoyed it. It put sound in. The effects are cool. I just I was very happy with this result. We also had we had the Jerry's apartment result. Now, this didn't necessarily look like Jerry's apartment, but if judged independently just as like a hey make a 3js interior scene, it was pretty cool. Some of the actual metallic effects on certain things like the light fixtures and even the sink handle and things. It was pretty cool the materials that used. Look at

the individual wood panel floors. That chair was definitely very odd. It had art on the wall. It did try to do the bike and that actual like rear bike tire is pretty interesting with the spokes and things like that. So, again, a like potential promising result. We also, in addition to that, how could we forget 3D printer sim. Now, I can't fully say that this is This is one where I basically said like don't think just output the code cuz we're having so many issues with actually getting it to output results. Fundamentally, this would have been good with given a few more tries. We can see it did actually have a moving spool, which is very, very rare to see. And beyond that, if we Let's slow this speed down. Let's try that one more time. We even see a little thing spinning here on the nozzle, which I don't think I even noticed the first time around. So, there were definitely elements here where it was like, "Huh, it really like did some due diligence." And that's very pleasing. The drum kit was very, very 2D. Though it was, I guess, photorealistic and the effects on the actual like hi-hat and stuff, as well as the auto play with the adjustable BPM, everything here worked fine and was nice. We did do the front-end test where it started with this weird like I hate this design aesthetic. So, I told it to change that and then put sample SVGs in if it does not have specific things that it can use for asset photos related to fashion. And the SVGs it put were actually kind of awesome. So, it did a good job, but it almost just kind of reskinned that same site, but kept that weird like vintage gold look that I've been seeing in some models lately, which I don't know how to feel about, but this was cool. The waxed utility pant. And then we had our low poly GP game, which needed some help with open code, but once we did that, it actually like it turned out okay. It had multiple camera views for first-person shooter. The bike models were not bad. The tires are actually spinning. It does do the wheelies. We had the opponent bike models as well. The um speedometer looks good. Okay, there's a Red Bull sign there that I did not did not notice the first time around. Shell, we have cool particle effects. We have a sand stands, trees, pyramids. This was not bad, and I kind of liked the low poly style, especially when it does actually do the wheelie effect. That was pretty cool to see. So, overall, the model definitely shows huge promise and potential. I was personally let down by the UX or user experience that I received from the AI Studio website, but again, the flip side of that is well, it's free. You have no right to complain, and they open-source the weight. So, I totally accept that and understand that, but I just wanted to really properly try to test this, and I still feel a little like incomplete in terms of being able to make a definitive judgment about this model. So, I'll be interested to see what other folks experience with this is as it does definitely seem like the potential exists. So, with that, that is going to conclude today's video. If you have any questions, please feel free to leave them in the comments, and thanks for watching.