Hello, everyone! We're on ToTheMoon. Technological news, Silicon Valley sites around the world. We're going out every Sunday. There are special releases on Wednesday. Sign up, by the way, who was watching our special edition with Sasha Mashapov on the last Wednesday. We've been through a new GPT read model, Antropic has been discussing, you know, different sites. And there's a lot of video generators, of course, especially given that Higgsfield is worth more than $1 billion. So, I liked the comment, "When that assessment is so interesting, Sasha has become more interesting to hear." That's the point!
Yeah, yeah, yeah.
Yeah, it's gonna be $10 billion worth everything, but it's like, "Of course, you should listen." All of it, already a lot...
You can also have English. I'm more clear.
It's a little less about English, I think.
"The winners are not sued" is called.
Yeah, yeah. Everything, everything, all right, that means. There's another type of comment. There's a lot of comments on it. Well, sign off who didn't see, look, and it's interesting to see the information. By the way, we're gonna-- we're gonna at least make a special edition, maybe it's just gonna be there in the night before the first day, yeah, where we're four, yeah, the thirty-first number is, the thirty-first. It's an environment where we, the four and our friend in English, so we'll tell you all the trends and all the details. You'll have to join us. Very useful information, cool, you know, especially in such a free, good format. So, the new autoAI pictures. How do you make an annex that goes to a top one in Android for 18 days with four people? OpenAI certification. We can get two more courses. OpenAI launched the first arts education courses and aims to teach, notice, 10 million by 2030. More than a large number of educational platforms. What happened to him in weeks? Australia has begun to apply the first law in the world to prohibit social media for adolescents under the age of 16. All teenagers. What status is it now? I suggest we proceed quickly on ChatGPT 5.2.
I'm here, Eldar, yesterday, and I sent what was in Pro's version... I'm glad to have this subject very much. So who has Pro theory, who, by the way, uses GPT chate or uses Ultra Gemini version, or uses Grok Heavy, the maximum version of Grok, for example. Write it down, please. It's very interesting to know who's got signatures, yes, so big. I use Pro. Tanya, Steve, Tanin's husband, we have a business with them, too. We're out of here together now, and I'm telling him, "So buy Pro quick." He was ready to buy me. He says, "Why do I need him?" I never got him to explain himself why he needs Pro because he uses a plus one percent, right? Well, I was on a plane with him, and he used it hard, and he had to.
He's been using the first two days all the time after he's been in touch with you. Then again, this wave is going to be quiet, and he never uses it.
He recently asked me if he had recently asked me how the GPT reads to synchronize with the post office and that the mails were automatically answered. I sent me how to do it, said, "I-I'm automatically letters to pee," right? Well, he's got a lot of people writing in real estate. Do I know what I did to him? I asked this question in GPT chat, threw a scrimp over him and wrote, "Take that question in GPT chat." Well, because I gave him a good answer. So he gave a super answer there. There are many...
You have Pro, and he's
got a plus.
No, I gave him a plus. I asked him a question on the plus side. I asked him a question in the usual thinking of the regime, didn't I? By the way, plus the version, there's thinking heavy. There's a Pro version of thinking and thinking hard. I think, uh, Pro's version is a regular plus version of it. That's what's a minus plus the story, that there's no way to choose this thinking, well, reasoning and to make a reasoning hard-core thinking. And by the way, I-- I also noticed that Steve, as a whole, probably a huge number of people who use GPT chat, so, a little, they're using, of course, automatic. And he's sitting, so he's getting an automatic answer. Says, "Something nonsense, some kind of nonsense." I say, "You're not a joke, you're not a waste of computers, so you're not wasting your power." And the question of capacity. So that's what I wanted to ask you. Those who use Pro, why couldn't I answer him? Because I'm Prousing a lot of things, and I'm launching big analysts there. And now, the very interesting analysts are when there is room for him to download more than 100,000, like phone calls, and he'll recognize it. Why can you just throw ZIP-Arhawk at it, and it's really incredible, right? So he's doing all the texts, and then, uh-- well, it's just a huge volume, and the analysis can be done for 30 minutes. There's a problem with Pro regime that Pro's thinking a lot, it's not a big deal. And now they've added a quick response in Pro mode and a sort of heavier response. I was just, Eldar, throwing you out yesterday because I didn't-- like I didn't see it, did I? I got him, too. Who tried Pro mode like this quick, fast-track response. I'm personally, by the way, here I am, in my reasoning, I've always got a strong reasoning, you know, in Pro. On the plus side, I don't think that's the thing. And Ilinar, you use hard, or--
Yeah, I'll get it out of here as soon as I get there.
Just a little bit of a stiff. Yeah, he's answering fast enough.
Maximum.
All the money.
But Pro doesn't always use you, right? Pro, you don't always use it.
More. No, well, most likely Pro.
More like Pro, right? Well, you're on professional matters or when you ask what difference it is, I don't know, there's a flower you're gonna ask for a simple question, too.
Illar is not asking you that.
Oh, Ilnar has questions only how to make a spaceship to fly to the moon or what?
Well, I really needed to, you know, help dad change the butlers for new ones. And I thought I was thinking, well, I'm gonna take a picture of it and ChatGPT Pro, make me the right answer, how you need to change the butlers right. There was some kind of design, something was needed. He finally thought something to me-- half an hour, and he gave me a huge instruction. I started reading it, I got lazy. I came in on YouTube and found a short video for twenty-eight seconds, of which the first ten seconds of a man is filming
an old palace.
That's enough.
Seventeen seconds. That's enough, yeah. I mean, many things are so domestic, simple--
It's a fact.
Write, viewers, what kind of cabs you have, where you're sure you'd better open up now, like YouTube or read the article in Google, and that's gonna be very interesting to hear. Or better, for example, at ChatGPT to find out. But I'm gonna agree with you, especially in the car, I changed some kind of filter, and-- a salon, air, and I opened it, I still found out with a GPT chat how to buy what to order. And he gave me that. He told me it was easy to change. I asked him. Usually, my attention to the car was a salon filter, and it's not a fact. And here I asked him to change easily. He wrote, "If there's a few filters in my car that change in one second. When I changed the air filter in 20 seconds, I thought why didn't I do it before? So now, of course, it's really fun, huh? I mean, with systems, but again, everything is on YouTube, yeah, everything is on YouTube. And the question here is, you're too much on YouTube, and it's hard to find. And write the briefcases. Anyway, the case is interesting to know.
Visually more understandable than the text is a huge read when you were all shown.
Yeah.
You saw that.
Well, it's fun here when he can, you know, it's fun when he can deliver the right pictures. I was asking him yesterday, for example, to find me big parks where I can go out with a dog without a lead. And he threw me some really cool places I didn't know at all. So he's, like, writing, like, 100-30-gactar, walk, there, no leash, there, in San Francisco. You're like, what? It's a huge amount of land, right? And he's putting pictures on, and he's got some photos on the spot.
It would be great if he added me more in detail, but again, I guess if I asked him.
It would be great if Ilnar had asked him to generode the video. I'd show him his butler.
Well, by the way, Gemini's gonna recognize the video and the audio. Google periodically, yes, periodically. Oh, Google, ChatGPT periodically. And I didn't get a lot of that, by the way. But I'm told that you feed him periodically, he'll recognize him periodically, but he's also a little bit of a hunch. He didn't want me to adjust PDF recently. He says, "I can't do it." Then you tell him, "You can do it. You just did it in another chat."
I believe you.
And he says, "Oh, well, then, yeah, I can do it."
And by the way, it's very interesting that we might be making a special release. I have a friend, a very strong AI-stor here. And now we're in touch with him--
Here. And now we talked about him, and he told me that he used a lot of mossies in the first place, not a resonance on many questions. Because there's one minus in the resoning, they didn't learn to clean it up. What's in the resoning, if there's a hallucination from the beginning, and it can move, not diminish, yes. And some things require that, sometimes, just a simple answer. I mean, so he's been sitting there for a long time, and he's been there a few days ago, why-- how much in the resoning might, why is there absolute and-idiotism in the resoning, yes, it's been issued? Well, because there's a hallucination on the hallucination, and there's a chance of error just beginning to grow dramatically. And he says they're still fighting it, they can't find it. And that's what it says to you like, on test rats, everything is being tested. Rabbits. Yeah. He says it's not an alpha version. As he said, it's very interesting, it's not alpha-version, it's, uh, purely research, purely research, as they are called. And they take you very well. He said rabbits first, then he said rats. He said it would be hamsters, he said. Then he says, "This is a fact," he says. It's just that we have to admit that what everyone's testing and what products are being plunged, why they're so ejected. That's very interesting, by the way. He had an interesting explanation. I asked him why the foods inside the good didn't show up, did they? Why did Sora let out, threw up, released the Image, there, the zaki-- released, dropped it.
They say, "There's a research organization, they're like real scholars and they're all going to be investigated." And he says, "The day will be released again." Let's see, by the way, we're out of a new Image, and we're going to go to him now. It's like a social network like that again, yes, the deposit is new. So you see some pictures of the generos. Remember, there was Sora? Sora's out. Sora's social network has emerged. What's wrong with her? Who uses it? Who comes in where the news is, where does all the information about her? You're stubborn, you're stubborn. We're waiting for a new Sora. Image is now released, a new deposit has been made. Everyone wrote how cool he was. It's obvious, like Elinar just said, the answer to Nana Banana, right?
Yeah. But the goal of all these releases, Sasha, is to support a lot of Haip. I mean, well, they let Sora out, maybe she doesn't use anyone. I'm not looking after the fate of Sora, the number of users, the activity, and so on, but they've collected their creams from it. So a lot of people are starting to talk about it again. The stock is probably on the rise. Attention, well, shares, no shares, because IPO is not here, but the attention of investors is probably also warmed up. About that, too, some benifites gathered in OpenAI and Gemini, and all the other companies that do. So, by producing unprepared, unproven products, they may be doing well.
So it's important for them to just keep a lot of memories on the market.
I guess. But what's the feeling now from the new version?
Yeah.
And when you said I remembered, 15-twenty years ago, the phrase was popular when you asked, you were told, "You were banished in Google? You don't know Gugle, do you?
And there was a separate reference, that's the answer. And you're asking some stupid question in the forum, and then there were forums. You get a link there, you're going through this reference, and there was a poster saying, "If you were given that reference, most likely the man who gave it to you, he thinks you weren't looking too closely at it. Google. You can ask your question in Google, and here you are, go over it there and ask your question."
I think something like this might well come up with a GPT chat now, when in response to some questions that might clearly be clear to him, there's a link like that.
But I can tell you, and I can honestly tell you, there's a story that goes on, where people didn't know, they need to explain and tell, and even more. I'm telling you. I didn't send Steve because I didn't want to explain because I realized he asked a very abstract question, and he'd be asking for a clarification in the GPT chat, and he'd ask for clarification. That's right, because even GPT chat says. He started instructions to give some people who I didn't even know. Yeah, I just don't want to be that specific case. It's just there's a lot of int-- these are now. In the section of integration, integration where it says yes, it's possible to connect different modules. They're not working. They're working, they're not working. I've been putting Google's calendar on, I've been meaning to do some chips to make him, uh, by the way, a very nice case. I have kids vacationing at school. I can, for example, ask the GPT chat room for a list of all weekends and holidays and holidays. He's texting me. I've been there before, he, I'm telling him, make me put them on my iPhone. He's making me up, I'm adding them to my iPhone. And even I'm so musical-- they sent concerts for my daughter. I said, "Oh, Elan, do you want me to give you a phone?" She says, "Oh, cool!" We couldn't give each other a hand. He didn't have a file on the iPhone. Oh, come on. So I decided to put him on my own calendar, because I'm on the phone with the module. Google Calendar is a module. There's a Photoshop, there's Adobe Acrobat Reader, there's a Booking, Expedia, well, a lot of stuff. And there's Canva, there, yeah, that's how we described it. And I wrote him, he couldn't do it. I mean, I terrorized him from all sides. He never got me in Google Calendar directly and I'm gonna add it to him because it's much more convenient than... Although it is relatively easier to click on one pre-established file and add a new calendar. I just didn't want to add a new calendar. Case, by the way, is really cool. Now, everyone who has kids or who's interested in holidays or holidays is a very nice case. So you're here, you're updating, GPT's chatting completely updates your whole weekend or those countries that need to be. You're asking for some extra conditions, he's putting them in your calendar. I don't care how you want, yes. It's a very cool case. And, uh-huh, again, so Steve didn't explain it. But I mean, many people want to actually send them to GPT chat 'cause they want, they spend their time, uh, as abstract, right? And sometimes people ask questions they don't... when they don't need a question. That's it. So, you got the same people, right? People ask questions just to ask.
That's why it's because of it, yeah.
Yeah, it's irrelevant to this case. I just remembered that 15, 20 years ago, that was a funny thing. Here. And with GPT chat, five two, and, yes, that's a clear answer. This code is red that was declared in OpenAI, it worked. They were fast-tracked by the drunkard, and they'd have finished their release and released as soon as possible. Uh, today, literally or there yesterday, at the time of the film, uh, Image Generator showed up with the next version. They've done a little update on the shore, haven't they? So they didn't make it second, they made it one-five. Still showing that it's not some crazy change, but it's a kind of workout. I've been throwing out the chat room, trying. I did what I did to the nano banana chat with the GPTschne new Image Generator. It's a good result. I mean, generally, uh, catch up where they're back, yes, and places, maybe even there's a little ahead. Uh, that's what we said, remember, when Gemini came out, that it's not necessary to jump off the OpenAI payroll to Gemini's payroll. Still, the plus-minus company is still in the short term, and it's not too good to know. There's no way someone ran ahead, and the other won't get it.
Yeah.
We've got to throw everything down there from one subscription to another. As for some professional tasks, uh, I haven't found a case yet, where I could definitely say that Gemini is a big cooler or a GPT five two are more cool. With the previous version, I thought Gemini was better. I'm not feeling a big difference right now. Yeah, that's why--
There is. You agree, you'll agree that there are some things in five two that have improved right now, right? I do, I do. I do. I was recently texted by Ilo, Ilo, Ilon Musk. Ilon Musk. So I'm writing, "Why are you calling Ilon Musk?" He says, "Well, I'm really wrong. But it's not a good thing to say that." I mean! He's writing it, that's all right. Well, I mean, I know there's a possibility that you're reading that from the word standpoint, but that person is still here. I thought about the neuron of Mask, too, with Elon Musk, there's a fight against them or not.
That's how they didn't solve the problem. She's not really in the way of it, but she still does, yeah, when you're in the middle of a depressor or when a long answer comes. That's probably the same situation you have when a part of the word is written in one language, a part of the word is written in another language. I have one, too, yeah. It's a very frequent phenomenon, because it's just thicken, they're conditionally related to the same thing. I mean, the same word, but it's just not, uh, that's not the kind of token that was needed for that language. I mean, this mistake is happening. It's a savage parole, and there's Chinese hieroglyphics coming up from time to time. GPT's chat, I see a part of the word in English periodically. You've got that, too.
So there's still a place to grow. It's still going on, but the parity plus-minus is restored. With the baccalaureate, there, you know, interesting, they're one of the baccalaureates just like they've been through. I mean, there's a double growth in indicators, and on the one hand, it looks promising, on the other hand, of course, that's undermining some trust. It's so strong. It feels like they just stumbled on this baccalaureate, and consequently it may have just become unrepresentative of these measurements. Here. But that's the details. How--
It's just that detail. Like, uh, Sasha, yeah, I was saying, uh, last year, when you and he were together, that for the domestic use of the baccalaureate, now, there's no more to show, because they're checking out, on a wide spectrum you never use in your life. But the same Humanity Last Exam, he's asking, there, on parole, about high math, about, there, some biology in the middle of something, there. Well, in the household, there, we ask how the butler is changing and sending a picture. Yeah, but what difference does it make with Humanity Last Exam? It's important how he answers your questions, which you most often torture him. So there's a pattern here. It's good to see her, see what models are moving, something else. But for domestic use, it's like not that big a thing.
Sometimes Elon Musk is gonna be a lot of groks going to some top bunny. You come in, and you're just gonna stop taking it at some point. I'm actually, like, I'm just starting to stop taking them. They always say, "We're cooler, we're cooler. No, we're cooler. We're in some kind of binchmarck. We're in the binchmarck." Now, uh, this is the-- the only thing I've noticed is that Grok, uh, Grok, uh, er, has e--e-e, has it. Elon Musk just reassigned that Grok was in the top, uh, top of the score, models, to-- to the top of the model speech to speech, yeah, right, voice, voice-over, voice-- where you speak, and what that is, Wait, there's, like, ninety-two points of three per cent, although Gemini has ninety-two points of one per cent. That's what it looks like, of course, interesting. And they're just about to get over the GPT chatting, like, 10 percentage points, right? And that's interesting to see what Grok did in a year. And here's what's good here? Look, there's a model that made this kind of a year. They're the ones who've been doing better, they're improving all the time, they're gonna go up there, they're gonna be all the time. They've released, well, they've got a pretty cool model. They've got , they have a speech to speak. By the way, they're saying that it's like XAI's estimate now, two hundred and thirty billion dollars. I thought, of course, Ilon Mack was a real man in terms of, uh, asset creation, right? In such a short time, it means that you're gonna snoop off a new asset like that, that's amazing.
Twenty-three billion.
And, yes, about the baccalaureate, what you're saying is that it's a very important and delicate story. Uh, we're on a lot of schedules. These baccoons are so much as you're even in memory, okay, five-- if you don't watch them actively, and you don't always say what's inside. And it's nice if you have to make a point of finding on the baccalaureate, you need to get in, see what kind of requests, the baccalaureate checks out what answers go. From this perspective, to look, and to what extent does it matter to you that the model is well performing these tasks. Remember, we used to give an example, how models are doing specific household tasks, as an example, there, a private investigator. From this point of view, you can come in and see what entry data you've got. Here are the episodes, I don't know, there's ten, well, ten, 15-twenty episodes back, we showed it. Accordingly, what input data give models that it episodes, as checked. From this point of view, you can see how important this baccalaureate is to you, right? And, in fact, all the benchmarks that are, they can be looked at if you're actually gonna make conclusions on the benchmarks. No, well, just look at the integrated meter, like some pretty pictures, and, in fact, you see someone else was going to beat somebody else again. What's the measure? Well, many people don't even care.
You don't have to be so big.
Yeah.
Oh, yeah. And about what you said, uh, thinking and extended thinking that appeared in Pro. What do you want to talk about here? Uh, since the appearance, there's GPT 2, well, 2, maybe nobody's been watching. GPT 3, 3.5 and 4. And that's all we know right now. First step what was it? The models grew up, their size increased. So there's GPT 3.5 more than GPT 3. GPT 4 is significantly higher than 3.5. There, there was 4.5. There's a big sa-sama, remember, right? Well, I don't remember, I think that GPT 4.5 came out this spring. It was the biggest monster size model that was. But she was not rizoning. So there was no rizoning first, and the quality of the model was increased, where the baccalaureate closes, how well they do, and so on, it was by increasing the size of the model. But at 4.5, GPT 4.5, this all stopped because it showed that the use of risoning when the model was about one, then the model was o3 was, much more tangible than the increase in the model size. And now we're moving to an increase with the size of the model, the length of these distracting chains, right? There's one point, I see, it looks good, but, roughly, the model is longer thinking and starting to answer better. And this process, it's not yet on the top of the border. And in the gPT Pro extended thinking, it's just a lot of time, more tokens to give the right answer. And on that basis, a better answer is possible. And in this sense, as if we're talking about, uh-a, GPT, there, Pro Version and Plus, Pro version is clearly winning not only by having a more qualitative model, it's got more tokens to think. Again, of course, you should always watch, and how real you need it. I mean, maybe Plus is totally there, with your head, closing all your needs, and there's no need to overpay. But in some kind of task, it makes sense to buy Pro a version to get more of this context window and time to think, the number of tokens that are being used by the model for thinking. Because, for example, the models that Gemini 3 Pro uses in Pro of the Google version, in Pro, rather, you know, not Google, but the Open in Pro. And the models that were used on mathematical olimpiads, taking five of six tasks, they spend a lot of tokens on the mind. They're not the only answer they're making, but they're getting a lot of answers. Then, based on this, the next model above determines, and which of these answers is better, it starts to be further developed. It's all been a waste of time, and it's all a waste of tokens.
So adding an extended thinking in Pro's version is just enough, uh, logical step, and it's definitely effective.
What do you think it was before? Before that, I think there was something mixed up. We didn't know what. We didn't know what level it was. I mean, now that's saying that Pro's theory, how cool is it, that you're adding different branches and then you're getting a choice. You can, like, get more power in there, right?
Yeah. We, we don't know how it was, we don't know how it is. But the trend is that it just increases the time to think and, in fact, it all starts to work better. It's today we were, uh, talking to the guys. Like we can, uh, put some versions in our service. One version will be light, the other will be extended, well, there, or there, thinking a version that spends more time. And one says, "If we give a thought to the version, everyone will only choose it. Why choose a better one when you can be better?" And we started laughing that we could just delay. Well, I mean, roughly, uh, the delayed answer. I mean, you're probably standing there, wait 15 seconds, just after that, start generating and sending it to the same model. Whatever. And the user won't understand the difference, actually. And there's no guarantee that the GPT chat has the same thing, there's no model like that, there's no algorithm. That when you choose to go on with your thinking, you have the same thing, and it's only 20 seconds later, right?
So you don't know how exactly the hood is, but the trend is that the increase in time, the increase in the number of tokens that are spent on reflection, results in a significant growth in the performances as yet. Models.
But it's the first-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a It's when I bought Grok Heavy, I remember, he just got out. It was just that you wrote that you started parallel, I think there are three or four agents, I don't remember, three or four. And they're just like they're getting in there. So you saw their lines, you saw the lines of those agents, right? And, uh, it's, uh, again, you see every time these agents are lined up, and you don't understand. It would be good if he gave you some sort of extra details. Because that's all the reasoning, and you're kind of in-- it's good to see how he's talking. It's not human, though, is it? I mean, it's not easy, it's not like you know it at all. I mean, he's, like, in Pro's version-
The right directions.
Yeah, Pro said, "I went three-ways." Well, of course, because when you're in Pro theories, you start deep research and you tell him, "Find me the news in a week," he's asking you, "This is the day?" And you're saying, "Get the table and just a list." He says, "I've got it right, you need to make a table and a general list, right?" And you're like that. Here. Well, I hope they'll, by the way, re-enact deep research, because they're deep research, I think they haven't updated for a year. I mean, I don't think so. But they tried to update him sometimes when they took out the questions, then two questions, and there was one question, right?
But it's some kind of nonsense. I'm just here to see if I can get here. We'll see. By the way, did you see multitasking in GPT chat? Multitasking button. Have you seen it?
No, I didn't.
So I got here. I've got some kind of stuff going on there. And there was a multitasking button. I didn't realize at first what it was. And she's...
I know what that is. And she looked like it was said, "Try to ask, there, try asking chat GPT something else while you wait." And, you know, while you're waiting, ask for GPT chat. I mean, I'm in the dialogue window, and it says, "Take GPT chat while you're waiting for something." I didn't see it, I didn't even see this anonna, and I asked myself about GPT, and he didn't know I had it. Very unwise. He said that, well, you know, you can ask the same dialogue window, he'll be kind of in charge. He'll say that the first request is still in the process. I thought it was weird, right? I mean, I asked him something, he started wasting his computer, and I asked him another. So he's better off compeling him. Well, I didn't realize how it was supposed to work in the same window, but they're testing, maybe there's one project that's probably testing this user's work in one window, which is, in general, Interesting, right? So we still have to come to the idea that this obviously switch between windows is such a very stupid format, yes, very stupid format. Not AI. He's obviously not AI. He's kind of working, okay, it's like folders, yeah, that's where they're separated. Projects, projects call it that. What projects? Well, you just put it somewhere. It's clearly some kind of format. I think it would be a long time ago that there would be some kind of a story, there, putting together all the different gategories, there, some sort of association, would have to build, connect you there for different chattels, I think it's a idea. But maybe we'll come to that someday. Still, the research organization, as my friend said, but they're obviously working on the topic of one window. And I think Ilnar and you were even talking a few years ago, that there's still work in the same window, yeah, at some point in time.
Well, at least we had this discussion. Doesn't mean it's true. Let's talk about the next one. So this is what the top of the app went to Android? Top one download. The app was tested for 18 days inside and over 10 days, right? Four chelas.
We've got a release, yes.
Yeah. What is important is that the annex was made in the Codex, it was done in the Codex and put it out as described how it happened. And that's an annex Sora. So, to date, the video generator for the OpenAI video has appeared on Android. It went right out there, I see, in the top. In addition, four persons have been listed in the Codex in the same way. And here comes a reminder of this company that Lovable, I think, is called.
Lovable, yes.
Yeah, which now has valued more than six billion, I think, if I'm not mistaken. She was just there the other day, she was on the news, and she had an evaluation, and this company, she's six and sixty-eight billion dollars in 13 months, she's attracted a round of three hundred and thirty million. I'm gonna need some money now. It's important that from zero to 100 million dollars, millions of dollars of proceeds, well, projected, it's been eight months, then between 100 and 200 in four months, and she's had an increase in her estimate over the past five months from one. Point eight to six points. But by the way, it's interesting to know that six months ago, the company was worth more than half a billion, right? It has been filled with over 100,000 applications, and it has created twenty-five million projects. And very serious, yes, very serious. And some kind of space-based, space-level. I mean, I don't know, I don't know, the details and everything, but still, I'll make a start-up worth six and six billion dollars. Okay, there's a lot of people doing such start-ups like John Ive, or Mira Murati, and that's another-- by the way, Jan Lee, Jan Lee Kun left Meta, yeah, and opened his start-up. You read that, right? And again, it's the idea that they all go away, open the start-up. Who works there? Here. And although models continue to be cool. And you can do it. Who did, by the way, write who's our audience. We do have, in fact, the developers are professional and unprofessional.
And write, interesting boxes, what you made mobile applications, what you did with the Codex or any other means of development.
Yeah, it's a very interesting story. So OpenAI released Sora under iPhone, under iOS, and accordingly decided that this would be possible to move her to Android. And tried to do that experiment. Four very cool producers were taken, and they were given access to the Codex accordingly. Well, they have access to Codex, but they just used it in a clear way. And there are some statistical indicators that are interesting. They spent five billion tokens there. At the bait, if they used it on API, it's $405,000. So you have a monthly salary for four developers plus $45,000 for the use of Codex. And, in fact, they were able to resolve the project a month from the beginning to the end. It's very interesting to read how they used the Codex. And here I am here to all people who are interested in the web code, who use it in practice, I strongly recommend that we go into the OpenAI blog and read it. And if it's short, it's the point. They tried to say, "We have the IOS code base here. Do the same, just Android." Well, I'll clean it up, but plus-minus, yeah, no extra stuff. And there was a button, and they started it. Aaaa-Pricula earned it, it got it, but it's bad. So she doesn't meet the safety requirements, the scale. They're not comfortable with this and they can't be released. I mean, just from the box, the model can't solve these complex cases yet, even the case files from the series, which you already have a ready-made code on Swift UI, and you need to move it from the sweater to the swift. - I'm not sure. Here are the programming languages for iOS and Android. And here's the guys show how they used it to make it work. They've been collecting all the foundations first, all the logic. Part of the code they used to do with their hands and said that, in a word, I need a new window. Here. And one of the windows of the line, I've already written here, just like that. And when they have already done such things, the Codex has become an acceptable result. They're just telling me what they're strong about, what they're not very strong about. But if it's short, it's the conclusion. The hard-core experts with him, with some experience in using Codex or any other system like them, can achieve very good results. Just write from the box, and make me that app so far, it's still not good. It may be someday, but as long as the normal sales of the raid, the app is almost unrealistic. Here. But I'm sending all our readers, just readers, readers, readers and read this article. Very interesting to describe how they did it. The only thing I want to celebrate. OpenAI like the owner of the Codex, and it obviously praises it. And that is, all the boxes that are given, you have to understand that, as a salesman of this system tells you. Just like when we discussed, remember, the Claude, oh, not Claude. The anthropists said we'd give Claude one team, and we'd never give another team. And there's gonna be some sort of task to do. And there's a Claude team doing it up there that's way better than a team without Claude. Again, we have to understand that Claude's maker tells you. Just like Suzkever says the current architecture has been running into the ceiling. We have to do scientific work on this subject. Well, as if he had a start-up who was doing scientific work on this track, he'd attract money to this story. And as much as these articles, all these statements should be viewed through the prize as if they were doing, right?
So it's very interesting to read, but still keep your mind in mind that, like the Creator of the Codex tells you about it.
Look, it's interesting that you've given an example. Many people can ask: why did OpenAI first make an application for Apple, not for Androids? And there's a very strong story. I mean, I think, uh, there's a feeling that, in the idea, Apple's share is less in the world than Android. And, indeed, Apple's share in the world is about 20 percent. The question is, in the US, the share of sixty per cent, the sixty-one per cent of Apple. And there's a case that reminds me of. I had a conversation a week ago, we were flying about-- talking about BlackBerry and Nokia, and that BlackBerry was like a super cool America thing, and so on. When I said, "But he wasn't a top one, there was nokia. I'm the one who looked at me and said, "What is Nokia?" And, in a series, even when I said nokia is a Finnish manufacturer, he said someone-- some country he called, I didn't even get it. Ah! He said, "Not Japan?" he said. And at this point of time, I realized something was wrong and I came into the data. And data, and I was very surprised at the data, because if BlackBerry was a 15-minute, 20 percent high, and Nokia was 40 percent in the world. At the peak, yes, in the 2,000-eighth year, Nokia was 40 per cent, and BlackBerry was there as much as two thousand-tenth year. At the peak, there's 15, 20 percent, and in the US, Nokia has about 5 to 10 percent, tops in the whole story, five to ten, just a little. BlackBerry's on.
A Blackberry reached forty-three per cent in-- for the period of two thousand tenths. Forty-three percent of the Blackberry market! Yeah, and that's very interesting that every country has its own way of taking any brands, yes, they hear. And the poet-- you just said they did for, uh, iPhone-- for Apple. And maybe just ask, why did Android do that, like, at the end? Someone's gonna say, "How's that? So...
For change.
A Sora was originally released only to the American market.
Yeah, yeah, yeah. And c-- just American. Yeah, and it's clear that OpenAI, uh, the main, well, the main audience that first needs to be captured is the American market. I mean, companies like, uh, Gemini, companies like OpenAI, companies like Grok, they're targeting, Meta, they're targeting a base initially on the American market. It is clear that they are already spreading to the world, like, for example, the Indian tariff is new, five dollars from OpenAI, right? So five dollars already came out. They were doing it for them, which is, like, ten dollars, right, already?
Oh, is that Google now?
Your models.
Google let me out.
Five dollars for India.
Yeah, they're just the same markets.
I saw the news, I-- like the OpenAI released all of this. They're just on the same-- on the same market. So, uh, yeah, a little education. I think history is very interesting in terms of, uh, what happens inside the projects in the world, which will be in education, yes. Uh, I can say, however, that now is the time when the education market finally has the opportunity to make, you know, fundamental changes to certain, they will be taking into account the emergence of AI, Probably close to AGI. But they'll do it. That is, in the next five years, these changes must occur. By the way, I wonder, two days ago, they gave information that the coursera bought Udemy, right? I was wondering why big operators wouldn't buy the coursera and Udemy and Udacity. They're buying Udemy. Here. But it's old companies, they're supposed to have nothing at all. What about, uh, what's OpenAI? OpenAI, first of all, has started an arts education course. And they said we wanted to teach ten million people by two thousand and three. Again, assuming that, as my friend modeled, that this research initiative would not work, it would be just the figures that were stated. But if we look at this initiative in general, even a little more research, they still say they want to teach ten million people of artificial intelligence. Ten million. Ten million people are decent. And very decent. And, uh, there, five years is a serious human flow, purely for the stub-- artificial intelligence. And, uh, OpenAI clearly wants to get in, uh, in there with his infrastructure, his paradigms, his own notions.
And ten million people are a serious thing. At the same time, I think we've been talking a little bit, but we haven't told you the details. I got the certification, didn't I, Elnar? Two programs. OpenAI certification is a very interesting subject, especially who is interested in obtaining different certificates or learning from you. Well, it's not the essence of age, I've said it. I mean, there's just people who are interested in getting a certificate.
There, uh, you're actually talking about the Coursera. E, at least one of the two courses, there, one basis and the second, in fact, the basis and for teachers, there, for use in the work. Here's the second course that teachers, uh, put him on the coursera OpenAI. It's actually a way to go through and there's a way to get a certification. I mean, it's not some kind of autoAI certification department. You get a diploma, there's a signed Sam Altman. It's still on the side platforms, but the OpenAI is moving well. Here. And there, on the same coursera, these products can be studied. At least for teachers, the exact coursera is a reference, there's a blog on this subject. Here. And interesting. On the other hand, of course, ten million people, which is not so ambitious when they have eight hundred million active users on a weekly basis.
Yeah, when they're planning to do two points six billion by two thousand in the 30s!
Yeah.
And I was gonna say that the idea is...
Users, yes.
Yeah.
A minimum billion people in school. But it looks like a research target.
Mm-hmm. It looks like something, yeah. I mean, they probably haven't been able to make a specific decision yet, but they're actually trying to find him. If they find targets, they'll probably become more ambitious.
I think that the guy who's just doing this project is, like, a big goal for him, like, everything. So whoever's inside is doing the project, this target looks big. Once again, just to teach ten million people, uh, artificial intelligence in general is a good story, yes. It's just, uh, if we talk about introducing mass-based education into the market and platforms, to create these new big platforms, it's not, of course, they don't-- they don't have that, at least, these, initiatives, right? Because there are things they're showing serious, well, they're saying seriously that we want two six billion users, yes, to have. Or we, there, we want to have hundreds of billions of dollars of money. I mean, like, things like that sound like that. They sound exactly how much they want to pay users on the first one to have what conversion will grow.
I don't remember how much they were, six percent of the paychecks said. Well, I don't remember, there was a number. Interest, right? And that, these things, at least they, uh, show up, yeah. And here, well, there's this., uh, graduation, I think four years ago, yeah, we were talking about a unique world-- and I would have drawn everyone's attention to this case. And a history. About a year ago, I remember, one man told me that Sasha, all that's going on in Australia is a test, test site for the United States of America. And he says, when something happens in Australia, then look back at the back of the other countries where it happens. And, well, I'm not really dealing with this, this, this, this, that, uh, inter-galic, inter-galic politics. But it's important that Australia has enacted a very serious law. I'm kind of close to him, he's not close to some side, yes, the fact of this law, because it's possible to fight forever. What does the law sound like? So I'll remind you who didn't see, see if you can also see that graduation, that Australia has started to apply the first law in the world prohibiting social networks for adolescents under the age of 16. Yeah? Kids. Once again, to make everyone understand what that means. Yes, children, yes! To make everyone understand what that means. It means that it's not just like someone in there who banned someone, that if there's any social network that's in there, there's no restriction on YouTube, for example, yes, but it's bound to TikTok there, there. limited to Meta there and so on. I mean, if, for example, by accident, Instagram doesn't close or close, and he doesn't do anything to keep the baby out of there, then the user's fine is $50,000, yes. So it's very serious. I mean, it's a very serious law. It's not a joke. Why is he in my whole neighborhood? Because just to fight to limit children in social media, I see myself that started using computers in the '90s, and I got two hundred and eighty-sixth Pentium. Well, I mean, yeah. He showed up at my house. When you have two hundred and eighty-sixth Pentium at your house- what old you are, Sasha. - Nine years, nine years. And parents, and parents don't understand what-- in the ninety-second, in ninety-second, I said, yes, yes, yes. Nine years, ninety-second year. And parents don't know what it is. I have a monitor like this, I have a sony monitor like this, I remember a big sony monitor. Which means, and, uh, of course, you're kind of putting it on computers, so... there's just something you could do. So there wasn't YouTube, there wasn't TikTok, there wasn't an Instagram. I couldn't go to the bathroom with this computer, shut up like that. Yeah, no one could come in, no one could turn it on, no one could check what they were doing. There was a command line, Norton Commander, there's a Windows, which should have been able to move a week, there and so on. I see. I see. Here. Prince of Persia and games, what games were there on the kidal. So. And the things we were getting into, came-- usually, we used to go down in one apartment at eight hours, and some of our parents came in and yelled at us. We were just leaving for another man's next door, so until no one came. But the point is, what is it? What time is this and given that I have four children, I realized that fighting social networks is an absolute, complete abstraction. I mean, it's a practically impossible design, provided you live in a society where children don't use social media. Well, that's how we have school, for example, we're not allowed to bring phones to school. So kids can't get a phone at school, they can't get it, they can't watch at all. I mean, no. So when you're not allowed to get a phone at school, that's not bad. I mean, logically, when you have a ban, a total ban, and at least a school setting, the baby doesn't have a phone, so... And when they say that no one can use it until 16, it's great because you go to school, you talk to your peers, you have friends, and you're not in it, and you're not in it, and you're in it.
This content doesn't consume.
And you don't use that content. I'm just saying, she's close to me, I'd be her, well, I feel like I haven't thought about her for a while. What do you think about this again? I think I'd be okay with that if that law came out, so that until 16 years, it would completely exclude all of this stuff, all of it. Yeah? So, but there's some good stuff, of course, but I'm gonna say that because I'm just at my place, you know, I was running a business that taught tens of thousands of children and they were sending kids to study. Well, it was like a by-product, right? Child education from six to fifteen years, right, Ilnar? Well, it was a little old. There were different programs. Ilnar was there, too, so he doesn't have to hide. So this one means--
But I didn't do any kind of
schooling.
Well, I didn't, yeah, I didn't do any of those kids. And there were different programs. And, yes, we're in child education, and we're not even in the same program, we haven't seen them, we've been very much with you, even including, we've been very much aware of adult education. And I often asked me why your kids were studying there or not? I'm telling my kids not to study because I don't want my baby coming, studying eight years at Roblox and, uh, shutting him up now. I wonder what it is now, well, in Russia, yes, it's been restricted. I wonder what happened to companies that sold Roblox courses. But the point is, you know what Roblox is really, what's going on inside, what kind of worlds is that, what, what, what, what, the kids are in? So it's sold under the programming initiative, and in reality it's just a toy. Yeah, that's the same thing. Okay, Illnar, two hundred eighty-sixth Pentium in eight-- ninety-second year, it was more programming than Roblox, because it had to be, oh, my God!
I had to turn it on.
There was a key, of course! So you should have, I remember, there's a-- I bought a processor. I remember burning Athlon, buying, so Athlon and bringing, he didn't put a puppet on him. Well, I just didn't know I had to put a culer on it. I put it back, said it doesn't work. I got a new one, but I knew I had to put a curler after that. I'm also such a fashion dealer to buy, I bought this big, big, big one. I liked it, so I've always had some interesting things, but I've been sleeping it because I didn't know where to put it. And, but the point is? What's the setting, like, Linux, huh? We were putting Linux, setting up, setting up Windows. Put Linux in the '90s, '90s, '39th, '94, 'cause I'm in a c-- or what year was it? Ninety-fifth, maybe, huh? Linux, I must have been ninety-fifth, ninety-sixth, so it's forever-- it's been a mix-up of that age. It's more serious programming, I think, and than Roblox programming. What do the audience think about this subject and what do they think about this law? This law, in fact, has nuances. One case was that law passed 20 years ago or even ten years ago. Another case came up with the law now that these teenagers were all, they were on the phone, they were with the instagrams, they were with the tictos, they were already with the-- TikTok, they were already restricted, too, right? YouTube's not limited, as I recall, bye. And they've already had a lot of other huge scoons. I imagine that such kids have lost these scouts. I mean, for the most part, which has been scattered by huge, big scouts, they've got these scouts. Yeah, it's a trauma. Yeah, it's a trauma, it's when I discussed Telegram locking, I read that one. So they asked, asked for one policy, and how do you block Telegram and Telegram, because Telegram, there are so many businesses running around there? People spent a lot of money making these scouts. He wrote, "This is not the question that should be thought of. It's a mistake." So, I guess, when you're in the process of introducing restrictions to 16 years, social media children, I think they had to consider this issue. Yet Australia, well, is quite on a lot of episodes, I think it's a democratic country, and they realize that they're going to cause a lot of stress, yes, in the country. I don't even know what I'm saying, but I hope it's gonna be okay with the kids. But for many kids, do you know what this is? Take their social networks and take away all their communication opportunities, all those details and everything.
Here.
Remember, we were talking about how they're gonna control it. Yeah, it's a condition, if there's just a spleen where you're your age-- you can just make it a little bit more and you can get a little bit more and a little bit more and you can get a little bit of a scout or someone else's, uh, a little bit of a scout, and it's a condition that someone's got a parents registering. The account is on it with its ID. And that's all right, kid, use it. But they've built systems that check the probable content, what the pages are going to be. You need to take a picture of it when you're authorizing, right? There's a selphie or something. And the interesting thing is, and it's actually tracking, right? And if, accordingly, the user's behaviour in this account suggests that this is a child, yes, who should not use it, the system makes a decision and locks it up there, yes, there until it is there There are no episodes to be resolved. Because, as you say, there's a $50,000 fine, a very big story, and companies are starting to track it. And there were funny facts that some children were able to go through the stage of the sponsorship. That's where the photo of the selphie was, snorting. So they start building some mushrooms. Yeah, yeah, yeah. And the model says, okay, you're like a lot more than that. Well, I mean, it's a model I'm telling you, the model can be hacked, you know, I mean, I can't find any of the boxes she didn't study. And there she probably didn't see the kids in the training that they're going to be. And they're there, snoring, taking pictures and passing through. I see they're gonna close up soon, right? If it's a big story, they'll be stuck with filters all the time. But this is the situation, and the companies are actively tracking themselves.
I remembered that. We've got a photo shoot here, and the man my producer, who takes pictures together, is filming, he says he's got a situation, so he looks very micro, very micro, very micro. And he tells me he had a situation when Gemini asked to do something, we wrote the kids without editing. He says I do. You know, we don't edit kids. I just remembered that means a case with that. I want to make a point about YouTube, by the way, because I've just over-checked some sources. I read in one source that YouTube is not in. YouTube is on the platform list. It works, it's there. There's a child's regime, I'm sure there is. He's in. There's no way that everything goes. Not even a child's regime. No, well, he owes me. I don't understand how it's like youTube down there when you have a lot of things in your house. But the essence and YouTube Kids is a child-regime, it doesn't count as a age restricted under these rules, does it? Here YouTube, but youTube Kids and safety are related to services that are not considered age restricted under these rules. So you're only allowed to enter YouTube from 16 plus years. Here. But again, how will they verify that? But what Eldar says, I just don't know how much they've spread this test around. Probably won't be a series of checks on my daughter in Instagram, and she's got a question there for how old are you? Well, you pick up any age and go on. I understand there's, like Eldar says, another type of verification is already under way. Yeah? Now, if parents, for example, have they switched something and given them, then parents will be punished for $550,000, too? Because there are parents who tell me no difference, take my phone and sit on my phone, right? Social networks. Is that how this will be regulated? The idea is that it's probably supposed to be a violation of the law for parents, too. No? I wonder if my parents are, yes, I'll write it now. If the parent gave the child a phone, it's spreading. Fifty million dollars. Well, I see. Straf. Yeah, I see. What kind of life? Look, but some things are actually meant to be forbidden, don't block it, don't fight it, just put in a big ticket. You're making a fine that if you have this app, you've got a million dollars fine, and they'll all wipe it fast. There's a man who's gonna think, you know, some activists.
Wait.
Yeah.
We're waiting for an explosion in the commentaries.
Yeah, we're waiting. Yeah, yeah, yeah. We're waiting for a bomb in the commentaries. But there are some.
Experts in all
fields.
Yeah, just from the point of view here, it's true, it's a really serious precedent. By the way, X is the X-File platform is Twitter. Yeah. I keep saying, it's Twitter because I think a lot of people are still, who are not, especially in the United States, who don't understand that X is Twitter. Interesting, by the way, now's the case. There are some people who have sue Twitter to use it for their start-up, allegedly a stiff brand that is not used. Yeah? I wonder what's going to happen to them. Who are they with? Did Mask decide to sue or what? So X supported these rules. So X is the last platform that agreed to comply with the law in Australia, and I don't know if you know or not that X has a huge number of trials in Europe and Xu have now issued a big fine, There, hundreds of millions of dollars. And until the US President, all the government line, a huge number of economists, businessmen, say what the Euro-Use is, blocking its own.
Which blocks freedom of speech and the mass of swings in Europe grew in parallel with the X. And at some point in time, X became number one news platform there, almost all the countries of Europe. So, you passed the phone, no exceptions, no restrictions on the platforms, no parents' fines, no children, no liability on the platforms. They must be reasonable steps to avoid an account. If a child simply watches a public content without entering the account, it's not collateral, it's allowed. If the app's phone is collateral to the account, even the parent, and the baby uses it, it looks like a round trip. Platforms are obliged to try to stop bypassing. Is that interesting? So she's gonna be scanning the baby's eyes all the time? Here. And...
Behaviour.
Call attention--
What he's watching.
Platforms are obliged to try to stop bypasses and can trigger a rechecking of signal restrictions. That means platforms are allowed to scan people. Yeah. So it's interesting. Yeah. I mean, if the account is collateral, they have to be completely with it. What if it's not bail? So I bought a TV, opened YouTube unfit, like a kid can watch everything or something?
Yeah, you can't publish your content. I mean, the meaning of social media when you open, there, your channel, there, your little buzzing or something--
No child content, Elnar, right? Oh, by the way...
In fact, how would, considering what content, if the kids are some normal, yeah, the legal content is publishing what's wrong with that? I think it's the opposite. The question is, what content they consume potentially dangerous. Because...
It looks like...
I think they're not gonna rule what's needed.
Or maybe Tanya, on the contrary, maybe here is exactly what Elnar said, they're interested in not showing up at all, and children are no longer in social life. nets and consume, and plus not to be created, including, i.e., not to be dependent on the
layouts.
All right, you're publishing your kids. Is that a child content? I'm, like, my kids all the time--
Yeah, but you have rules for the publication of a child content. You know, yeah, you can't publish naked sysies.
No, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, Don't give me that. But...
You can publish a grown-up child. No, I can go to the gym with the naked. No? I can lay down.
You're not a woman.
But I'm not doing this. Look, I'm not gonna. In America, America, America, America, America, for what you just said, in America, what you said, it doesn't work, does it? About humiliation. So what do I mean? I don't think that's the law, is it? Because this is the haun. So I'm not gonna do this, am I? But I think, as I recall, I may be aware that the audience will write, in Instagram, I can put myself out without a shirt on the beach, and now the baby without a shirt, I can't put on the beach, right? Even if he's three years old and he's running there. I remember listening to a lot of cabs, a lot of lock-ups, and I always felt very careful about putting you out, so you wouldn't get banged by accident, right? Since I put a lot of content on social media, it was important for me to, for example, not get me. Here. And maybe it's one, too, of those stories. Although, for example, it is still considered that the content of the child is not disappearing from the law. The Act on the Prohibition of Agunts to 16, not at the disposal of the content. And you can watch public content without the entrance, yes, and use it to a lesser extent.
You can't watch Instagram without the entrance, I think it's the first time there is.
No, the fact remains that you have kids for-- you love to watch other kids too. So if we look at the trends, then a lot of-- kids aren't adults watching, they don't watch the scoons we're watching, but they can watch, I don't know, dogs running around like that, right? But no, not a content in terms of, uh, youth scouts. So they're looking at each other. This is, however, the second part, which is clearly, obviously in the area of political interest. We'll watch what happens to it. Very interesting. Although, of course, this case is him, he-- it's gonna take time. If that country had not used it in the first place, there might have been some conclusions. And that's gonna take time. But all people who live in big countries of another type don't worry, yes, in your countries, it's not likely to happen. Although we live in a world when I think the United States of America--
There's something else that can happen.
Yeah. Maybe something else.
So worry.
Worried. And I'll see you in a week. Togumud Canal.