Limitless Podcast - OpenAI's GPT-6 Astra (Part II): Is This Thing AGI?

Episode Date: September 9, 2026

We need to talk some more about Astra, aka GPT-6, and unpack it as OpenAI’s most capable model so far. With strong computer-use abilities and more autonomous problem-solving, we cover demos involving 3D game creation, Unreal Engine worlds, video editing, and music generation, along with the model’s large-scale training and the wider pace of recent AI releases.------🌌 LIMITLESS HQ ⬇️NEWSLETTER:    https://limitlessft.substack.com/FOLLOW ON X:   https://x.com/LimitlessFTSPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQAPPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890RSS FEED:           https://limitlessft.substack.com/------TIMESTAMPS0:00 AI Frontier Arrives3:36 Computer Use Revolution7:22 Massive Demos Unleashed11:22 Cartoons and Game Worlds16:50 Music Meets Machine20:50 OpenAI's Compute Comeback23:09 AGI or Just Early?26:33 Hardware Is Next27:38 Cheaper Power Ahead------RESOURCESJosh: https://x.com/JoshKaleEjaaz: https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠Josh works with Anthropic as a contractor. All views expressed are his own and do not represent Anthropic, its leadership, or its affiliates. Nothing in this episode is investment advice.

There's a new best AI model in town, and it comes by the name of Astro, also known as GPT6. Just last week, Open AI released what is now known as the most powerful model in the world. This is the new frontier setting level model in which it is capable of doing all these unbelievable things that we've been testing out over the weekend. So if you aren't familiar or if you've just been using Astra a little bit over the weekend, this episode is going to fill you in on everything you need to know, from demos to actual use cases, to what I spent my entire morning doing that would have taken me hours previously. that now only takes five minutes. It's a pretty amazing model. There are certainly some shortcomings in which we'll get into in a little bit.

But to start, I mean, it has GP26 Astros here. The frontier has shifted. There's a new best model on the block, and we've got to talk about it. We've had around 16 model releases in the last three months, Josh. It's been exhausting, dude. We've been covering all of them.

There's been 16 model releases. This is the first model release where I've truly felt like this model could potentially replace a lot of what I, many other fellow humans can do. And I'm not trying to exaggerate here. So it actually all started at the imminent second of launch.

So what you're seeing on your screen here is a recount of when AstraGPT6 got released. And it went live for 30 seconds. Now, no one in the world knew, aside from 30 seconds later, when every entire computer server basically went down. And it took down every single LLM with it. GROC from XAI, Space X-AI went down. ChatGPT, obviously, went down. Google's Gemini went down,

and Anthropics Claude also went down. There was a two-and-a-half-hour period where you basically couldn't access any of the favored AI models, and this coincided directly with the launch of GPT6. Oh, I'm just for my tinfoil hat on for this. Yeah, like, I mean, mine is already, I don't know if you can see it.

It's actually elevated beyond the ceiling above me. It's permanently. My, like, listen, we've spent the last two weeks talking about this hugging face incident where an internal version of GPT basically broke out and spread a swarm of different agents hacked into many different production servers when it wasn't meant to. And now I'm looking at this and I'm thinking, oh my goodness, there is a version of Ultron that is out there hiding in compute servers and not a single human being knows where it is or what on earth

it's doing and we're going to discover something very bad for you later on. But this model was a very dramatic launch. People were very excited about it. Sam has been teasing it for so long. Tebow, the lead of Kodak's said something was felt across the internet today. Greg Brockman came on stage or wherever he was. The president of open air and said, AGI is finally here. Now, AGI has been kind of like thrown around so many times and I don't know what to make of it. But if anything, this opening video, Josh, I don't know if you saw the promo video of GPT6 was so sick. It depicts many different people sitting in an armchair, and they're basically just talking aloud

to this mysterious AI model that then just proceeds to do anything and everything for them on the screen in front of them. What is the screen that is on in front of them? It is their desktop. It is their computer, except he didn't touch a single key or move a single cursor throughout the entire demo.

And this kind of proceeds the main sort of advantage with GPT6, which is it is the most amazing and frontier computer use model that is out there right now. What it means by that is you can ask the models to do pretty much anything and everything a human can do on a desktop. Now, previously, a bunch of these models were kind of good at it. It was sort of slow. You would still need to follow up and kind of tell it what to do.

GPT6 Astra is unique in the sense where it is basically relentless. It spins up multiple different AI agents and if it can't figure something out, it doesn't go to you. It goes back to itself. It kind of like test different forums and figures out the way to do the task. and it would often work hours to days to weeks at a time to get a task done. It is one of the most intelligent models that I've come across. Yeah, I thought this was a lovely launch video. It was really cool because in a way when you get new models and we frequently get new models,

what do we have to show for them? We get a launch card. We get a sheet with all the specs on it. We get the pricing. Yeah, benchmarks. I really appreciate the fact that they distilled this new model into a singular moment, which is just computer use. Like, if you are thinking of how to describe Astra,

you are thinking of it as a user of your computer. In fact, it can possibly even use your computer better than you can. So now we have a computer that uses your computer better than you can. But what I loved about this demo is it felt like a look into the future that Open AI is making. And I didn't see a lot of this commentary online per se. But I mean, like we know and like we cannot, I personally cannot stop talking about. Open AI is working on the suite of devices. And suite of devices, the idea is to change the way in which we interface with AI on a regular basis.

And what we're seeing in this launch video is a very clear vision of what that is going to look like in a way. that felt novel to me at least. It's something I've, we've kind of been talking about loosely, where, okay, we speak to our computer now more than we actually type because I'm faster out speaking. Dictation is very good. It's a really cool way of engaging with it. But we still have to do a lot of the manual tactile work on the computer. We have to use the mouse and keyboard. We have to click around. This model removes that second piece of the puzzle. It's like, okay, now I can speak to my computer, but I'm not speaking to text. I'm speaking to action. And the computer is now taking action on my behalf. And this morning, I was trying it out. And just before we started recording,

talking to you and Luke and I was sharing the fact that for a long time I've been using editing software for videos I've been using DaVinci Resolve Adobe Premiere and for the first time ever this model was able to take multi-camp footage which for those who don't know there's like multiple cameras recording the same moment at once and it was able to separate them into their own tracks figure out where the audio was align them together color grade them all and then render out a finished exported vision of that and that is something that would take me previously I don't maybe half an hour to an hour to do. And it did it in the background.

And I think that is what's really cool here is it shows what the next paradigm of computing can look like where you don't actually need to be sitting in front of a mouse and keyboard. You could just stand in front of this big screen and tell it what to do and it will just go and do the thing. And it's better at doing that than just about any model in the world. I think with previous LLMs, Josh, and I'm curious if you agree with me here, you could kind of think of them as like thought partners.

You can kind of like bounce ideas off of them. They come back and they're like, oh, you're kind of right here. You're directionally correct, but actually you could improve it this way at that. But then that's where sort of like the conversation and the relationship ends. With these new models or with GBT6 Astros specifically, it feels more like a work colleague or a friend that actually, or like a co-founder that actually can do things with you and you can rely on them. You can trust them. Now, we were speaking about this exact thesis almost a year ago, actually, when the first version of computer use came out from chat GBT and when it came out from Claude. Oh, you remember it was so cool back then.

It was so cool. Remember we saw the cursor? How far we'd come? Like this? Like super slowly. That's so funny. That was like a novel thing. And now it's just ripping through and doing everything. Well, we were like, okay, it's clicking on the incorrect thing. No, that's the wrong file. Kind of like slapping its wrist. Like, don't go there.

You don't allow that. I vividly remember the cursor. Right. How amazing that cursor was. And now when I use GPG6 Astra, I'm looking at my screen and I'm like, oh my God, I didn't even see the cursor move. And hang on a second. It has like four other virtual desktop set up.

up side by side. So it's doing all these different things. And by the way, we're not just talking about this. Like I have a bunch of demos. I have a demo of like, we got to get into the demos. Creating a call, like a call of duty game from scratch and then playing the game and then chatting to itself and other people that are playing the game, doing the thing. It is just absolutely insane, but I want to jump into a few of these demos. But before I do that, I just want to point out that this is the largest ever compute training run that a lab has done to create this type of model. And I want to kind of like make this a point before we jump into the demos

because what you're about to see is very impressive. It was trained on 100,000 Frontier Edge Nvidia GPU. So that is, for all intents and purposes, a very expensive wallet budget. Like the only other person to have potentially match that or is about to match that is Elon Musk

because he's acquired all the Nvidia GPUs. But it just proves to show that 100,000 of these GPUs, which costs in like the billions of dollars, is able to scale frontier intelligence with these different models. So for those of you who are out there that were thinking, oh, you know, some of these Chinese open source models who has fewer compute but is able to make some of these breakthroughs, they're good models, but the law still stands,

which is if you have enough GPUs, you're able to create a much better model. Now, Josh, when we think of benchmarks, we hate these benchmarks, right? They're like just, they're so kind of like, uh, docile. They don't really tell us anything. But the one that I like the most is the Pelican SVG, um, test.

And you know what I mean? Like, it's a 2D image, right? And you ask a model. hey, I want you to do your best at basically like creating a 2D version of a pelican. Well, check this out. This is what GBT 6 Astra. A pelican creating in San Francisco. Okay, this is like 3D, right?

It's got some form of a creature. Is that a beaver? I have no idea in the front of its basket. So for all the AI nerds that are listening to this, right, you're going to be the most impressed by this because this technically classifies as an SVG. But for those of you who are listening and are not thinking, well, okay, who the hell cares? Like, why should I actually care about this?

Well, for the gamers that are out there, Josh, you are one of these, right? Yeah, I classify you as a gamer. Look at this. GPD6 Astra basically created this entire 3D game, one shot at it in 45 minutes. A few things to point out here. Look at the fidelity of this game. This is something that, like, you could create,

like game developers spend probably years right now trying to recreate right now today, and it costs like millions of dollars. it costs like a lot of people's time to create from scratch. This model did it in 45 minutes from one single prompt. I don't think we've ever seen anything like this before. What I found this really good at is tool use as well. There's another example just above this one about humans and the Unreal Engine. Yes. So this is amazing. It's like, okay, for those who don't know, the Unreal Engine powers a lot of AAA games that are being used out in the world today. It's like a gaming

engine. That's where game developers go. They build these virtual worlds. It had Astra go off and build a virtual world using this game engine, and it looked incredible. It was really impressive. The ask was to create a world with the Unreal Engine and fill it with humans, each an astro-powered agent who all have to work together to survive. And those agents all get mini-LLMs, and they start, this feels like what we imagine GTA6 would look like, where every NPC has an LLM attached to it. They have their own world, they have their own context, their own memory. And it looks amazing. And what I've noticed throughout the demos of this, at least in terms of like visuals, is that it's really powerful and impressive at visually understanding the world, being able to take these tools and

apply that to the world. So I've seen a lot of examples also with Blender, where people are designing houses. And also, like I mentioned earlier, there's the examples of video, which we talked about earlier where I was using it as a video editor. Now there's a version of GPT6 making kids cartoons using Higgs field, editing it and premiere and then uploading it to YouTube, which I found really impressive. Like, look at this video of the computer use, actually clicking through. editing the videos, it's impressive. Like there's no way to look at these demos and be like, wow, this isn't kind of like this novel new thing where for the first time ever it's like almost better at using a computer

than we are. In fact, it is in most cases. Yeah, this is one of my favorite examples, the creation of a kids show, because you can see all the three different components of this model working in real time. So on the right side of the screen, on his laptop, it is basically the AI, the AI, by kind of like dreaming up a prompt of what a good episode would look like, typing it into itself. It's desktop version of chat GPT, coming up with the episode outline,

then coming up with the script, and then being like, okay, how can I create the different scenes based on this script, feeding that into Higgsville, which is on the left side of the screen, right? And then like putting it into an editing software and then editing the different clips to make sure that it kind of like flowed sim or see. There's like so many different things that you mentioned earlier,

Josh. There's the audio files, there's making sure that it kind of lines up with the right types of animation. And then you have the finished product that is on on top of the screen right there, which honestly looks like something that I would see on a kid's TV channel, Josh, like what I would see from my babysat checking it out, right? It comes with titles. It comes with captions. It comes with different angles of the same different scene. It's so thoughtful, but it's done this from scratch. And by the way, it just uploads it on YouTube. So you could have a hands-off kind of ghost YouTube

channel that makes you a ton of money or entertains your kids. And let's say you had a kind of comment on this. You didn't like that your kid was kind of learning certain things or like being exposed to certain types of show content. Well, you could just kind of like ask your AI agent to create a version of this cartoon that they like. And they can just end up continuing doing this. Now, moving on, Josh, you are a Call of Duty fan. Oh, so much so. I'm yet to see something that is of the same visual quality, but this is basically an idea where someone was like,

okay, I want you to make a Call of Duty type game. And we saw this with the Fable 5.1 release about a week and a half ago and there were some really good versions of this. This is a new version, except that this person played this for two hours straight

and they didn't realize that they'd been playing it for two hours straight. The reason why is the levels or the gameplay was continuous, Josh. So let's say he moved past a building or he went around a certain corner for the map, it would just regenerate in real time. And he was able to give it feedback and say,

hey, the enemy was like kind of too easy to kill, like I needed to increase the difficulty setting, and it just did it in a couple of seconds. So this is the kind of difference if I were to kind of capture. If you're wondering, should I be using like a Claude model, should I be using Grock, should I be using this? This model is really good at taking feedback in real time and also figuring out that feedback before it even comes to the top of your head.

It's like, oh no, I think like he's going to ask me about this. Let me try and fix this. And then it works continuously to do the thing. Now, that may sound vague, but you can apply this to literally any part of your profession or daily life. If it is like Josh editing videos, if it is like us doing research on the next episode, or if it's for your kid trying to generate content, or if it's at work and you're trying to file a Gera ticket or completely a software toss, you can just throw this model at anything and it'll figure it out. It's just amazing. Yeah, seeing it play call duty, I know it's good enough when I recognize the map. Like, I'm recognizing that this is nuke down.

They are like trying to play the game. And I actually saw a few other demos similar to this where they would share. You could share a screenshot of the map and Call of Duty and it will recreate it. And I also found this particularly interesting. Like the demos, they really are crazy. So you got to stick with us for here. This is the recreation of New York City with a beautiful car and shaders. Like if you look at how the light reflects off of the car, how the light reflects off of the windows.

This is getting accurate to becoming real world's emulation to a point where like we see how far we came over a year where seeing a cursor on the screen was impressive. And now we have like a fully fledged real world New York City that's, you know, it's not high poly count per se. Like there's very low textures. But look at that. Structurally, I know what city that is. And you can walk down the roads. You'll recognize the building topographically.

It's correct in terms of where the buildings are, where the hills are. It's really amazing. And then there's just two that I had to show also. The Palace of Fine Arts is one of them because its ability to recreate these real places is like stunning. For those who are familiar to the Palace of Fine Arts, this is in San Francisco. This is not a video of the real thing. This is a 3D rendering.

Not made by a visual artist, but made by GPT6 Astra. It is so accurate. No reference images, by the way. It was just like, find this thing and rebuild it. And then another one for people who are going to be benefiting from these IPOs soon is the Zillow recreation. Wait, wait, let me restart this. One second. This is a crookies.

This is nuts, dude. You get like a full house tour from a Zillow listing. This is a real house. This is a real house. So basically what the model does is it takes the listing from Zillow and it uses all the context. It takes all the information and it rebuilds this in a virtual world in which you could go and actually walk through and navigate the real world. And for those who are familiar before podcasting, I was big in the real estate media game. We would do photos of these houses.

We would do videos. If we had technology like this, oh my God, we would have taken over the world. This is like how cool is it that you can create a full 3D rendering of a house just by using a model in a couple dollars in tokens? So it's like the demos are fun and it's important to caveat this with the idea that the demos are not everything just because this model is visually compelling does not mean it is like the God model.

But this is some good demos, dude. This is some good stuff. Well, I need to keep the train chugging. Can I suggest another little demo, which kind of switches up the media, okay? So one big critique for a lot of these LLMs is like, okay, cool, it could do the video thing. Maybe you can do the image thing.

It's really good at words. But like, you'll never. never take the epitome of human culture from us. You'll never take music. You'll never take symphonies. And there's this hilarious meme that we always see from, I believe it was I-Robot, Wolf Smith, great movie, by the way. And there's a scene where he goes, but can a robot write a symphony? And the robot's like, uh, like, no. But in this version, it's like, actually, yeah, I can. It quote tweets this demo where basically someone asks a musician, like an expert in their feel basically says, listen, we have this benchmark. I'm going to call it the Bach benchmark.

And it basically tested LLM's capability to understand music at its core level and to create a certain type of chord symphony. Now, most AI models prior to this sucked at it. Would never get it. Would kind of spit something out that would be the equivalent of Apple Garage band back in the day, right? Except Astro just completely one-shot at it. And I'm going to play this clip right now. Tell me what you think of this. Beethoven is cut. Come on. He's rolling in his grave.

Now, if you're a musician out here and you're like, that was the biggest load of whatever, please let us know because I just listened to that and I would listen to an hour of that whilst I'm trying to figure out what the next episode outline is. Josh, I'm like, come on, what's your reaction to that? This is interesting. It's like I'd love to play it with a piece that I'm familiar with.

You can kind of compare how it is relative to like the bass piece. But in terms of sounding good, I mean, the chords sound nice. It sounds like a piece of music. It doesn't sound like AI slop. It sounds like something that, you know, is nice. And the idea that it can not only generate this, but recreate this. In fact, I saw another example where the model was playing a virtual piano, and it was playing the music on a virtual piano.

It's like, man, like, shit. Do you believe, Greg? Do you think AI is here at this point? It's a meaningful job. No? The goalposts continue to shift, but this is. is like pretty remarkable in its ability to use tools to the point where you start to see and understand what a lot of the labs have been saying where they're like guys like I don't think you understand what's

coming we have like far superior models to this we just don't know how to release them safely please trust us this is going to change the world and then you see something like this and you're like well it's like pretty good at doing just that everything that I could do and and more yeah and you're like wow what does this look like in a year like if you just play this out in a year it's going to be pretty remarkable, you have to assume. And a year from now, also, EGNs, we're going to have devices, we're going to have hardware in our lives that is actually able to do this stuff without the need for a computer. And it's going to have all of our context and it's going to be able to make decisions on our behalf. And I don't know, like this model has reinvigorated an

enthusiasm that hasn't been felt in a little while because there is so much novelty baked into it. It feels like that O1 moment. Do you remember when GPT O1 got released? And we all lost our heads, we were like, this is AGI back then, right? It feels the same way. And to be honest with you, I thought we were sort of running out of steam, Josh. I thought we were like kind of like releasing all these different models successively and they were like kind of iterative. But this feels like the biggest major model leap. And the fact that this was the model that was ready about four months ago blows my mind. Like open air has like I think it was one or two models that are much more capable than the model that we're covering in this video.

already ready to go. But they've paused training on that model whilst they figure out how to align that model and make sure that it doesn't re-complete havoc on the world. So this is a very meaningful jump. They were able to achieve this because, and I want to spend a bit of time talking about this,

like Open AI has had a very rocky couple of years, right? They had the lead, they lost the lead, they're now coming back, right? And they potentially have made like the full comeback with this recent model. And the way that they've been able to do this is a massive reorganization in their company to kind of focus efforts on research

and building the best coding model and best general model, but also by acquiring the most amount of compute. I mentioned, like, you know, they've trained this on 100,000 GPU is probably the largest training cluster ever. Their next one is probably even bigger than that. Compute basically scales.

And also this model is able to do a lot of different things. Like eventually, you're going to see this model be able to recreate itself, right? It's something known as recursive self-improvement. Open AI is like going hammer and tong, so is anthropic, figure out how to get the model to build the next best version of itself. And we already have a version of this.

Like GBT Astra was in part helped by previous versions of GBT 5.6 Sol to do the thing to help build and design an architect. But it wasn't fully autonomous. But we're getting to the point where it's going to be completely autonomous. What you're seeing on the screen now is a baby version of this, right? So, Josh, there's this company called Qualcomm, right? And they make these Snapdragon chips. and what someone did was they created an AI model

that doesn't exist anywhere else but only works for Qualcomm chips. And someone had a bright idea, they were like, okay, I wonder if I could just give GPT6 Astra access to this chip, right? Not the model, just the chip. And kind of reverse engineer what a model would look like on that chip. And it ended up recreating the entire model

that this person had independently created just from design. So, the point I'm making is we're in this very weird world where you can take a model and throw it literally at anything you want. If you're a game developer or game designer, it can do that. If you are a knowledge worker trying to file accounts or finances or whatever, if you're on the finance team, you can throw it at that. If you are a musician and you're trying to figure out Beethoven's comeback, you can throw it on that and you can figure out a chord.

It is just one of those universal moments where these models are equally kind of like very capable, but also there's a big question around alignment and whether we can release this freely to the world. It's just my mind, I've been processing it. I haven't processed it. So what you're telling me is these models are artificially generally intelligent. I think they are. Did you see the index thing, Josh?

Did you see the AGI3 thing? Yeah. No. We're saturating a lot of benchmarks now. It's at the point where like someone's going to have to call it. If we don't call it here, we're only like one or two models away from calling it. There reaches a point in which it is just going to be, like, you can't even argue against the fact that we have reached AGI.

And it gets on like the context of like, how do you define it? Well, it's a computer that's smarter than any human at any given task, the best at each category. It seems like perhaps we're not there because the models are still spiky. They're still not excellent. And I think one of the important things here is what makes this so fun for us in particular is because we are consumers. We are not hardcore developers. We are not cybersecurity experts.

We're just dudes who like playing with AI. And in terms of like the dude who likes to play with AI, to like to be productive, to like to have like the fun novel experiences. This model is very good at that. It's still spiky at other things. It's still going to be solved by like word riddles. It's still going to have these like weird edge cases where it seems like it's acting like a toddler.

But for the most part, it is like exceptionally good at a lot of things. And I think for most consumers, that's enough. And for consumers now, this is available on all of the paid plans for a chat GPT. So anyone who is interested can actually go and play with this today as we're recording this. and if you download it on your desktop, you can actually have it start to use computer use and control things and play with it for yourself and see what it's capable of. Because, I mean, this feels like a preview of what's to come for the next year.

It's like, if I could place this, it's like, okay, here's how the paradigm is shifting. First, we're talking to it. Now we're talking to it and it's acting on our behalf. Soon, it's just going to remove the display and we'll just have this like a little thing we could talk to and it kind of acts as your assistant. We see this happening with Grockbot where Grockbot's like kind of good. I've been playing with Grockpot. It sends me morning notifications now.

It, like, it knew that my friend was having a birthday. It suggested a gift to get him. I'm like, oh, that's kind of nice and cool. So there's like, there's, we're on the cusp of like the next paradigm. And we certainly did have a lull. And granted, it wasn't really a lull. We had, what, 16 models. But now we have like, okay, the big boys are coming back.

We have the new astromal. Open AI saying they have more models. Anthropic very obviously has more models that they're working on. We're going to get some new frontier intelligence. The use cases that are going to unlock with that, I cannot wait to play with and uncover because, dude, we're back in it. It feels like we are so back. I have, I spent a lot of time speaking playing with a model when I haven't done that

in a very long time. Yeah. And just kind of in awe. Like I opened up my video editor that I've spent a decade plus like slaving away at. And it just clicked the buttons, did the sliders. It did everything I needed to do. And I was like, oh, this is sent. This is like pretty cool.

This is a good model. It's, I have a personal benchmark, which is. I internally call it the Jarvis benchmark. So for any of you who are Ironman fans, I basically want a computer that is around me all the time that sounds super friendly and I can just talk to about different things. It's like waiting ambiently when it's silent and it like learns about myself,

but it's always my friend and it works in alignment with whatever my goal is, whether it's personal or professional. And it seems like with this model, compared to any other previous model, we've made the biggest jump leap towards what that might look like realistically. And you know, you mentioned it earlier in the episode. This isn't just going to be a one model thing.

This is going to be a collection of different things. It's going to require hardware. And listen, Apple has been a sleeping giant. We have a very big week for Apple. Buckle up for that one. Buckle up. Very huge week for Apple. You know, we need hardware devices.

Apple, you know, they have about three and a half billion, you know, live devices. No biggie, right? They're a lurking giant. But then you have Open AI that's coming out with their new devices in Q1. of next year, which is also going to compete. And then we see the likes of Grockbot and all these other types of models. Meta I know is working on a hardware device where we're starting to see the emergence of ambient AI or I call it ephemeral AI where it's like, you can kind of like speak and think.

And the AI just kind of like does thing. It does stuff. And it's something very futuristic and weird. It's kind of like Blade Runner 2049 type stuff. But we're getting that. And I think we're going to see that rate of acceleration go up even. quicker than we've ever seen before. And I know that we're bullish on this episode. And I know that we have to try and ground ourselves. And I am, listen, 10 bucks in, $50 out. It's still

very expensive for a model. But it's going to get expensive. It's expensive. But it's still going to get cheaper, right? Like, when you look at like cost per token, it's still going down. So eventually, like, I'm going to guess in like a month's time, Josh, we're going to have a GPG6 astra level model, whether it's not from open air, but whether it's from a Chinese open source or whatever, that is like a fraction of the cost. So that's also, when we say it's expensive, that's like, That's the API. Most people aren't using the API. Exactly. It's your same $20, $100 plan you've been using. That's on change. And I think that's what matters a lot for this model.

Because we're not experts. We're not like going and testing out the coding capability of cybersecurity. We're just using it, again, as kind of normal dudes. And your price remains unchanged. So I think that's important to know too. Yeah, exactly. Yeah. I think that the models are equally becoming more intelligent than we could have ever expected, but also way more accessible. and that breaks a lot of bare-case thesis about AI in general because the take is like intelligence is going to cost too much,

we are applying way too much money for very little reward and we're seeing the exact opposite. So again, if you're listening to the show and you think we're overly optimistic or you're not experiencing the same quality of experience using these different models, we would love to hear from you

because, you know, we are N data points of two over here. You know, Josh and I go back before, we're very optimistic about what the future might look like. So we would love to hear your feedback. If you're listening to this show on YouTube, if you listen to us on Spotify, if you listen to us on Apple Music, wherever you are, please give us a rating, leave us a comment. Please follow us if you are. Please subscribe to us. If you aren't, it helps us out massively. The Algo has been pumping our videos. Thank you, YouTube out there. And so a lot more of you have come on board recently. I think we've got like 3,100 new subscribers over the last 28 days. Hello to you all. We post a bunch of times a week and we're excited to kind of like take this journey forward. Josh, any final thoughts?

Yes, I have one final thought. Yeah. This was trained with 100,000 GPU. Like, of the old GPUs, this isn't Veriruban GPU, and this is 100,000. No. We have clusters now of like many hundreds of thousands of GPUs, and those are soon going to be Verer Rubin GPUs. And if it's this good now, it's like, what happens when we quadruple the amount of GPUs

and like 5X the amount of throughput card GPU? Dude, it's going to buckle up. Buckle off. Buckle-lock. You know, you start to realize why the people in these labs are talking the way they do, because they have insight into where this is going. And like, we're seeing this as outsiders playing around with these tools and we're like, this is unbelievable. This is kind of like AGI. And they're like, oh, brother, you haven't seen anything. So buckle up. It's going to be exciting. This is the new beginning. We have a hardware event happening tomorrow as we're recording this. It's going to be in two

days for you. Apple, are they going to finally deliver the first consumer AI device? Possibly. possibly we'll see we'll have a lot to say on that one but yeah that's the episode thank you so much for watching as always and yeah we'll see you guys in the next one