Hello, everyone! We're on ToTheMoon. Technological news, Silicon Valley sites and the world. I think it's a very important week. An important week, in particular, that the Anthropic Fable and the Mythos family returned, the world's most cool model today. Although ChatGPT produces five points six and a new model family in parallel. A unique, uh, unique story, which, as long as no one in reality has access. And there's information that Google is delaying the graduation, uh, his model. And today we will talk about this, because, of course, now is the time of understanding what is going on in the market and what models are really, really, really cool and what we're going through in the coming days and months.
Uh, Tanya's profile is closest to the target audience of this service. I was wondering if Tanya was trying because you and I are sitting in the carpet. Now the news is Claude Science showed up, yes, I think they called him. That's a science environment. So they have a coward environment, such as a programmer, for this use of a home, even if you can say, right? There's an environment for Claude Design designers. And accordingly, now on the side of scientists, they have also expanded this story. Claude Science added. Here.
Come on. Very interesting.
I wonder.
work at Claude Code or Claude Cowork?
Cowork is more comfortable. Uh, maybe it sounds a little bit unwise, that it's not in the terminal that writes and all. I don't know.
Yeah. Yeah, but there's chat rooms, there's a deposit box, there's a code. Well, I'm sitting in the deposit code, I'm sitting. And there, there's no place, and there's no terminal.
Yeah. I'm probably the one who uses the same thing. Here.
I see her for some reason, many people call her a carpet, even though she's Claude Code for me. I've heard that many times. They were talking about the terminal. I say, "I'm not sitting in a terminal in any way."
Here. And Claude Code, Sasha, it's term utilisation. So you have to open this black screen and put Claude in it. Then you start, and you're like a hacker in movies writing everything in a terminal, all the slash-based teams, and all that. Here. But yes, I use UI and Codex, and Claude. I'm totally fine. I think it's more comfortable than, uh, a term story. I know the terminals, but it's easier for me here, and it's more common to do than write in the terminal.
Yeah, a little bit of that. I agree.
The graphic design is bigger, animation.
Of course! He's not a web, but, yeah, it's a graphic design.
Yeah, yeah, yeah. But I still want to study because no one did that before.
Sure.
I mean, uh, uh... The story that the carpet is called with, there, programming, access to files and stuff, there, is one thing. The second thing is design. Still, a little UI they change, and you're still on API to the same service. It's just that there may be more visualization, a little more stubborn. And now they've added Claude Science. There are, uh, well, all the examples in fact, which are given in this product's card, they're more biologically researched. I understand that it grew up more from there. Here. But it's interesting to try. I hope that a little math time will be running him off this weekend, and that colleagues will have some relevant work and talk to Claude Science on this subject. Maybe it's interesting. Anyway, if I do, I'll tell you if I've been able to get back to math a little while, although I've been five or six years of science. Here. Tanya, you'll be interested in hearing what's about design. Do you think there's something interesting to do? What's his taste? Can he create something more or less harmonious?
Well, you know, it's interesting that
everyone always writes that this tool and don't expect it to taste. That it's the only thing that no LLM has yet is taste. Well, let's see. Interesting. I'm really glad. Thank you, Illar, for sharing.
All right, we have a lot of comments, Tanya, where they say, "Get the real safer than Tanya has done what you're suggesting."
Look, I'm done, I'm giving up and I'm on the internment course. So wait.
I'm gonna say, I'm gonna say now that I see different courses on the creation of agents, you know, on the side. And on the one hand, of course, if you get to the interesting people, you can do some cool things. And on the other hand, there's a huge problem, of course, because a large number of people in the world are now parasited that they, uh... As they can. Because of the artificial intelligence that came, they became like everything, we figured out this topic, which they're helping to instil artificial intelligence. Look what happened before. There were consulting companies that helped businesses grow up there or were companies that supplied them to the software. Now, all those who have access to artificial intelligence, they've become both programmers and consuals. Yeah, I see these offers that are spreading everywhere. So we'll build you everything, and we'll help you in business. You say, "Do you understand how to give advice here? You're actually really into this?" We had one person who wrote in the commentaries that what I did was a total blank and easy thing. I wrote to him, "I'm hiring $100,000 a month for work. If it's really easy for you, you know, if it's one percent of your knowledge, it's an absolute element, then you'll make $100 million a month." He's really gone. There are no comments. And...
He's looking for your email, Sasha.
And there's an email hanging. We've got emails. People write things to us periodically, people write things to us. It's just that it's important, I think, very much, very important, that everyone around here is helping, everyone's learning, everybody's getting built. I just wanted to say that on the one hand, it's good for someone to go, turn, and do something, and study something. On the other hand, here I am, personally, I have always come to the same idea that the intelligence agents can only be dealt with every day by trying to use it in their routines. Life, ChatGPT, et cetera. I mean, I saw a lot of people doing training prospects. For example, as in ChatGPT requests to write or how to write in Gemini. Even today, I got a post, there was, don't use that word in AI systems, never say "I don't agree." Don't ever say "Take it," but say it. And you're sitting there, and you're sitting, and you're... That's what a joke is. Why?
I mean, do you think you're--
I've used all those words. What am I supposed to wait for now?
Okay, Tanya, look...
The cars' discovery.
Look, that's the misunderstanding of all this story. Because on that side, they make intelligence that should understand what you mean when you write no to him. It's you when you wrote "no." And Greg Brockmann was from OpenAI, he wrote something or, well, someone from OpenAI, they somehow said that, on the contrary, less try to give too much as a powerful one over the installations and try to program. system, limit it. I had to do it two years ago. Two years ago, you really had to do that when you wrote the system, give her a lot of different installations, that's the right description of everything. But we are now going to an era ahead of it, where it is not necessary. Yes, Elnar? This is very pure in Claude there's 4.8 Opus, there, or ChatGPT 5.5. You know, like you don't describe many things, but you don't describe many things, and you're more likely to get the result than many things you're trying to, well, it's very cruel to give in those terms, So the words that were right. It's obvious that this isn't all the case files, not all the time, and there's a lot to learn. But I'm just calling all of this and-- you, Tanya, with agents, you, Tanya, with agents.
Sasha, not just agents. You know, by the way, you've raised an interesting subject. We've been seeing my friend recently. She's a pretty successful lawyer, but she's also trying to put AI into her business as far as possible, and she's doing it well. She told me where to study. But she said the interesting thing is, she says, "I mean, at least half an hour a day now, "to communicate with," I don't know what you're using, ChatGPT or something to make him better recognize you. I mean, you know, like you're kids-- de-
How you spend your time with the kids, yes, your day, your husband, your wife, friends, your parents, someone. She said that it was like talking to GPT chat, so she'd be putting more information, some, uh, joint thoughts, asking questions. So you think it's not real anymore, that you're giving less information, the more it can be, uh, interesting output,
yeah.
I think that's a little different. How long you spend, how much information you give there. I believe that when a person spends 30 minutes each day strictly on something, he, uh, is still some student of this process, not real life. When you're really using this in your life, you have no idea 30 minutes a day or 15 hours a day, or two minutes a day. Because you use artificial intelligence as necessary in your life. Just as we use cell phone or smartphone, yeah, we open when we need a search system, when we need to watch pictures when we need to, we don't know, open some kind of cell phone. The document is when we need to open some mail and so on. So, to date, this tool is just an additional tool for your life, just like the phone. And say that, imagine, mobile phones have appeared, and you say, to understand the cell phone, you have to devote 30 minutes a day. But it sounds like absurd now, but it was, in fact, before. I've got a computer in two thousand-- oh, 1,000, nine hundred and ninety-second, you know, about that, in those years, yeah. Internet, I don't know, ninety-sixth or some year. And I remember very clearly, well, from the ere, there, I don't know, 2,000 fifth to 2,000 seventh year, when a huge number of people said, "Here you are on the computer, so you have to sit down, learn to use--
Well, that's really what I should have.
Acute, Exelem, et cetera. But the fun is that Exel's people learned to use it not because they went to study separately, not because they went to Excel courses, and Axel learned how to use people and how to use them. Use the same as people can print. But you can print a blind print like that. But Exel and Ward, if you start using it in your life when you use it all the time, you create it inside, make different formulas, do different reports, and plan budgets, watch. I'm sure you'll learn to use Excel. If you're in Excel, I'm thinking again about my wife, who's not working on desktop, never once in her life, yes, she doesn't know what it is. And if she'd be there periodically, like, she'd understand how she works, I don't know, notes, yeah, or how the M.O. works. Now, the SMSki is a fast-tracking man, so she can either lick a tape in Instagram or, like, get a registration on some website somewhere, so, yeah. Here are tickets, uh, no way to buy a trip, right? Four buttons, and then buy a dress on the website, ow, fill a basket easy, huh?
And the liki. So I don't down here and increase, by the way. That's the point. Where does that skill come from?
Who needs
it?
I mean, you use-- I, and I call, so I don't have that story to limit. What I've said now is a more fundamental question. And from the point of view, if you don't use all these systems all day, if you don't work with agents very easily, like the phone, with the Esemes, in the sense that you want to. If you do something, what you don't need, what you don't care, it's not gonna do anything more than that in today's world. And sometimes it might be some good, but it's a university thing to remember. Why did I teach LISP programming at university? Yeah, it's such tapes.
Or I taught Latin too.
And here you can say I taught LISP. I have been in the twenty programming languages to form some foundation for engineering, common ground. It was such a time. You taught Latin. Latin, by the way, that's a 100% plus. And Latin, I would now recommend that people learn. We have a common friend who recently explained to me the need for Latin knowledge. But...
You ask him to explain it to me, too. I still don't understand.
But it's both useful and useful, and the usefulness of it, yes. So, uh, but learn how to set yourself up, I don't know, some micro thing that generates the image is a simple rule. Igor came to me yesterday and asked me to buy Pokemon a card with a picture of us. I'm going to pay for it on Amazon, and I'm writing, you're gonna pay two dollars for being animated, even though it was written in the first place. I understand that I don't care about two dollars there, but I thought it was weird. I opened the GPT chat, we were shot and wrote, "you need anim. It did. I say, ow, this is animosity. Can I have another, more? It made me. I say, oh, that's fun. Igor, do you like it? He says, yes, it's awesome. I put it in, and she went. I realized they would have made animosity not that it was fun because it, for example, changed a little structure. And Igor has long hair, it's from him, like a girl, but she's actually probably thinking. He says, as I'm like Ivette, as my sister is like a sister. But he might have thought in the picture like that, and he automatically, in this anim, would have worked a little bit. Well, he painted her face, actually, he added it, and he painted it, like, how he swelled her face, right? And look, this one, this-- why did I open GPT chat? Well, sweat-- I don't do animos ever. I'm not making pictures, I'm gonna be honest with you. I don't know, this is who we have in the audience, just the people who use this service. There's someone who doesn't use it at all. But for me, as much as GPT is a simple real tool, that I've completely wrapped up this case and abso-- and easily used it. But for that, I don't need to study the V--- the pictures. Well, well, it's-- Elnar, remember when we Midjourney were the first, remember when Midjourney tried to make complex images from the spiritual world there? And that's where you should have been able to describe, right? Well, it must have been a time, yes, Elnar, when, uh, it was really hard to describe the proms, but now, as a little world has changed. Doesn't mean you don't need to describe the prom.
Look, no way, no-- I'm making a lot of things at GPT 5 chat, a whole archive of documents, proms, there, details and everything.
Sas, what you're talking about, you can describe a little hypertrophiced metaphor. An example of that. Look, you're here to see a doctor and a hot-- and, well, you're not feeling well, you want him to treat you. And don't tell him you're just feeling bad and you have symptoms like that, and tell him, "Let me measure the temperature." Well, anyway, he can measure your temperature. So you're too much of a special effort to give him the specific actions he needs to do, not describe the problem that needs to be solved. If you are not an expert in this field yourself, then perhaps you just don't, don't see the solution that was needed, and you're sending this instrument the wrong way. Or you come to a consult with a lawyer and don't describe the problem that needs to be resolved, what happened to you, right? And you say the specific actions that this lawyer needs to do. If you're aware of this, you know, you're an expert, and so on, it's a normal way. But when you limit yourself and you're not an expert in this part, there may be problems here that, Sas, you're describing. Therefore, for powerful models, some episodes need not be given too, uh, narrow instructions. If you're not sure you have to narrow down the course of his actions.
So, Darinar, I totally agree with you that not giving instructions, but more contexts like giving him a life in the first place. Because, Sas, even as you complained all the time, he makes some stupid recommendations to you, what you do on the weekends. So he doesn't know you as a person, he knows more about what you do at work, what kind of requests you have, the main employees who buy, compare it.
I have a lot of personality, but you're right. The question is, how he's still aggregating them, Tanya. And of course, he-- he doesn't even know my job. Well, let's be honest. He recently told me that--
He reads so much as that, yes, no
wonder.
Of course! That's a problem. Look, this is the problem that it is produced, is directed towards certain types or individuals, or to a certain nature, or to certain knowledge, to certain methodologies. Here we go-- we got a special release two or three days ago. I was talking about a very interesting study that was done by a large group of people there. Washington Post made a micro-study. What did they do? They asked thirty questions, uh, politically directed, left and right, about, uh, you know, whether the death penalty was necessary, there, right? Or should there be programmes of different sexes, there, and, well, so on. We need taxes on the rich. So they were assigned to different systems, there, Gemini, Grok, and so on. And you can see how the system responds differently. Do you know how many left questions, the ultra-field answers from the OpenAI? Eighty percent, and the right one is zero. Everything else, I think, is medium. Gemini, for example, has no left or right, only a mid-point. Well, he gives both. He always avoids conflict and gives both.
Something neutral, huh?
And it shows-- yes, and here in-- here in, you know, it's left and right. What's this split? And even there's a question, why did some questions call left, ultra-right or ultra-right? It was there to ask if to attack another country. Well, look at the specials. He seems very interesting. He's a little totally, very cool. But the point is, what-- what are people? People are different, they don't share the same kind of left and right, and they don't share only boys and girls. Yeah, we all have an incredible big life. And I see when he analyzed my chat, I still haven't given up this project, yes, we're all in the canal. I'll give it to him at some point. I'm gonna have to analyze these five or some chat rooms, over there. The thing is, he-- that was a problem. What he's been doing in some places like raising a business has made it important for him. I said, "It's nothing. I've been doing some research on a piece somewhere." Or said I had a nationality somewhere in a country. I said, "I didn't know that man at all." Or said, "Alexandre is interested in Egypt." I said, "I didn't ask Egypt. I just asked about eight hundred questions because I went there in March for three weeks."
Martha, I went there for three weeks a month, and I was wondering, and now I'm not interested. And then he found a separate blog where he was interested in some of the av-authores in Egypt with certain spiritual knowledge. He says, "This is a big interest." Straight. He's not a tourist interest and traveling anymore, he's already built interest. That I study and read literature and tie it to the fucking Egypt. Well, what I asked him in two years or in three years is an incredible number of questions about other books. I asked him, studied, analyzed, he kind of ignored it or privatized something. That's the moment the system doesn't see all this volume. And plus the system has some opinions, right? And therefore, returning to this subject, the same way we discussed the courses, you imagine that you're dating another person who also says some of his opinions, some of your knowledge latitude is still in the midst of your knowledge. Some systems. Because the systems are different. Because look at this issue, where is the percentage distribution of Grock, for example, left and right answers.
Well, though, I thought they were gonna be a little different, yeah, a little different. And, well, it's gonna be interesting to see. Let's go to the most important theme. We've got a break! So, a breakthrough, I think, in the world, is a model. And I still have a mystery for me. Is that marketing, fairy, what's going on? Because I think everything has become completely unruly, ununderstood and unknown. So what happened? Well, not the kind of thing that's happened to micro, though, I think it's the super important news. Claude let me in, Antropic opened his Fable model. It's available until July 7.
Sasha, it's important to make a note. I hope she wasn't banished by the time that this roll came out.
They won't shut her down!
Or something's not gonna happen to her.
Yeah, they won't shut her down! She's because she opened up last night, right? Or, at the time of the filming of that roll. I mean, we're in fresh-- we're in fresh things, and maybe we could just... take off the tape, say it's still not open, and it's already open, it's closed, right? And then she's available. It says she's available until July 7. Nobody knows what that means. She'll sell separately for money, she's still like this, and so on. Speaking of, uh, eating thickens. There was a fun subject. Someone of those famous programmers, I don't remember who's exactly.
I saw X show the schedule, and it says Fable crests of thicken. So, at the beginning of the Fable eating thicken, then Opus four eight meals of thokane, then there's ChatGPT five, just five, a little bit less token. Although I don't understand it, and how they measured it. Because literally this week, there was a theme when ChatGPT five was eating, Codex eaten at a large number of people in the tokens in just a second. And then it turns out it was a mistake. He's got them all over. And, accordingly, I don't understand how they're doing. I mean, who thinks and understands how agents work, like where. I understand that there's more tokens worth, right? Which indicates how much the entrance is worth, the exit. But I remember there's a mark that says modern systems like Fable or like four eight points Opus, or there's ChatGPT five high. Compared to that, I don't know, with the new graduated Sonet five that Claude released, they might be, they're--they're worth more, but they're more efficient and less toxic. Is that right, Ildar?
Yeah, and there's a baccalaureate here that thinks the dollar cost a job.
A specific task.
Yeah. We take the task, we put it in different models. They spend different amounts of tokens, and these tokens are different from the different models. And we can all make it money. I mean, Fable's on the $550 target. Opus four eight, for example, there, $30 or more. Yeah, and it could be different--
Can I ask you a question? Can I ask you a question?
Yeah, sure.
What if I threw a task and something that Opus four-eight has decided to suck, and Fable has decided to make it, unlike four-eight.
Yeah, yeah, yeah. So you're looking at it first, whether it's solved or not, or how much it cost. And if she didn't decide, you'd still spend money, right? It's like this here, too.
Yeah, there are tasks they don't know or see. Although this is interesting, what you're saying.
Sasha, do you know what's interesting? Sonet's five out, right? We'll talk about it, too. They call him a working horse. But in fact, on some of the tasks of this Sonet baccalaureate, five are more expensive than Fable. I mean, it's just a little shocked. But as much as Sonet five more...
I saw it when Fable got out two weeks ago.
Yeah, yeah, yeah. More money is spent on the task than Fable, which seems to be expensive and fattening to the thickness, not in itself.
Yeah. And there was a subject like that. Remember, in terms of quality of the task. Sometimes, models that seem a little worse in another mode, they can do better because they don't leave, I don't know, in some place, some kind of super-fancy. Although I am personally, it is clear that now four points are eight Opus Ultra Code, ChatGPT five points, Pro. All right, I'm out of the way because ChatGPT has taken this rizing off. There's a high, extra high. I'm just, I'm already out of pony-oh, thinking, thinking. I've already stopped understanding what that means.
I don't know.
I'm leaving myself...
I hang out every time.
And the man won't save...
How am I higher or higher?
Of course! Okay, there's a human-- and there's a Pro, and there's a Pro extended even.
Yeah.
And there's Pro as usual.
How is that?
And if-- and of course you're surprised that you're making a system for billions of people and giving people that kind of thing. But, okay, they do, we're hamsters, hamsters, but this is the subject, it's interesting.
So, Fable's open, see how much it's available. In fact, what's interesting in the Fable so you all know. Like, if you're programming in Fable, Cloud Code, you have limits. If you're in Fable, for example, sitting in the maximum version and asking him in the chat room, there's no limit, right?
Well, there's no limit, right? Just like ChatGPT has five points Pro. So, interesting things to see. For example, sometimes in terms of architecture analysis, LAN infrastructure. Ildar, right? You can unload all the repositories there or some part of the code and send it to chat, by the idea. It's called Claude Fable Max, it's gonna be there, right? It's not gonna be Ultra Code and, well, it's the ones with the top version of it that don't have the maximum version. Okay, go to the other one and you can analyze it there and it can give. You can save the tokens, you can calculate certain tasks, you can do an analysis and proceed with the four eight Opus or, for example, Sonnet five, which Illnar said. But before Sonnet five, I still want to say, GPT's got five points six. Oh, uh, it says it's just space and there's a schedule. I hope we show him here. Where's five points six, it means it's Fable, and it's like the Mythos family, which is, like, this is a very serious thing. What did I get to notice? I was confused by five points six because five four, five, five, six, five, five, five, five, six. I think if something really comes out, it should be called six zeros or not missing a n-- number. It's supposed to be that Apple at some point in time is gonna have to remove the number from the iPhone and at some point in time, it's gonna be there, like, just iPhone. But we don't know if it's gonna happen or it won't. Let's see how this is going to be. Well, here's the GPT chat at some point in time, I think, I should take this version out. So the story should be gone. But why would you want it? It's like now, actually, it's called Fable, Fable Five. Well, you know it's probably five zeros, and the other one. And you are, and Opus is four-eight. And you're what's going on? And why can't you just leave Opus, clean up four eight, leave Fable just. Okay.
A little shipping just about the name, since we started talking. Just why Fable, huh? Here, Tanya, I don't know, you know, you don't. They are different size models. And there's Haiku, Sonnet, Opus and now Fable. Well, Haiku is in Russian, actually, a hoku. Little Japanese poems, actually. Sonnet is a little more literary. Opus is a large literary work. Fable is a strike. I think the strike is no more Opus, but somehow. So they went on the literature to that end. Here. I mean, Fable is like a continuation of this literary history.
Look, this subject, what you're saying, is good, I mean, why else have a number, it's probably some kind of temporary solution.
We'll see. And here's GPT chatting, making a move he hasn't had yet. He says the model is called not GPT, five points six Pro, for example, is it? And she blows the brain on people, I think it's blowing up now. The following is that the model is what is actually called Soul Flagman. There's a cheaper Terra model and there's a fast Luna. And I like that word at all. There's still a quick one. It's so fast, it's a quality, quality, fast, it's fast-track. And then it's coming. Maybe, but maybe a flagship fast, so it shows up- and this fast Luna comes along. And nobody knows what it is. Why? Because OpenAI says, "Well, we've all released it, but no one's gonna get access to it because we've given the State the test, and we're gonna get some small-sized company, yeah, some kind of thing. "Absolutely selected small companies." You're so And that's why, of course, it's still unrecognizable. And on the market, there's a lot of discussion on the market. Some people say these models are really cool, and some people say that it's really bullshit. Not that it's a total, not a complete thing, it's not a big deal, is it? That it's not like a space that separates something. Speaking of Fable, we don't understand the whole thing about space or some kind of hack. Yeah? Because if you take four points, eight Opus, compared to one year ago, so four points eight Opus are alien space co--
It's an alien spaceship with aliens, well, there, I don't know, yeah, with aliens, there, and with some unique weapon and everything. If you take it, see what she's really doing. And I still like people in the commentaries who write, "Actual intelligence is a fairy, it's like a brick.
So he can't do anything, he can generate words." There's always someone like that. Here, uh, just like I'm called humane. Illar, you, you know, what's your vision of five points six?
Look, uh, baccoons they showed, well, they said we weren't showing all the baccalaureates, just because it's a bonus and so on. But they showed themselves in a good way. But, of course, the model row will drive everyone crazy because there is this five six Sol, which is exactly the one to choose, I don't know, there, the depth of reasoning, maybe there's something inside there. Then this Terra, she'll have her set of parameters, and the Moon will have its set of parameters, too. So, uh, how much? A year and a half ago, you said it would be good for them to simplify the model, and at some point they started simplifying it. Now, I think the zoo is gonna be all over, and nobody's gonna figure it out. Uh, like you and I said, yes, ChatGPT, OpenAI, I mean, there was something we should have said to the myths, so, the fairies. We did. But until we see these models, we can't feel them, at least at the time of the launch, at the time of the release. When a full release with a rolling card on everyone, I think it's a little more detailed, but it's good that the answer happened.
We're waiting for the same answer from Google. I hope they don't get too much in there.
Yeah, but for people, so everyone can know about Goodl-Gugla. Google-- it was expected that Google would release a new version of the whole thing. She was supposed to be out now. Here. And she was supposed to be out in June, as I understand, not now, but June. And, yes, it was June. And it says Google stopped the graduation, so it won't be released now. And they were supposed to release 3.5 Pro, discussed that they would release Pro, and they were supposed to be not ready for her release because of some infrastructure stuff. We don't know you, maybe they're not ready for the quality of this model. Maybe they're afraid there's something going on with the government. I can't believe they have a model now, at least, in the sense of Gemini moving, that the model is really up to the OpenAI and Anthropic levels now, in terms of these complex systems, especially the development. Because the fact that the Fable is famous, I think, and Mythos, it's still systems that aren't just getting a chat response, right? It's just, it's not even, it's not just a system for processing some analytical information, it's still systems that can create something or find some kind of vulnerability. For example, it says even the system, there, the Opus has developed a code for people who hacked some site ticket and started to give each other tickets. There was a case.
Started to give each other all the tickets. I understand the problem-- we understand, yes, the problem is that it's impossible to limit the system. So she'll find a lot of vulnerabilities anyway, and a lot of companies will get hurt. In particular, with regard to Google and the infrastructure, they were interested in the news that Google restricted access, uh, to limit the use of its Gemini systems inside Meta, which is supposed to be Meta. start consuming. Well, we know, yeah, we were talking about this, that Meta, there, trillion toxins of billions of dollars, that used so much data that Meta's demand for this Gemini surpassed that Google could have provided without. damage to other clients. And that, by the way, I think it's a big, big, sure, question, who's the system to give, what infrastructure to allocate. And the question is, we all worked for these systems for free. Well, there are people who think they're making money on us, that they're buying our money, and so on. But look at the costs of these companies, their profits. And I know we're all super-loss. What do you think about Gemini? What do you think about the Sol? What do you think about, uh, all this market now, an endless model market?
So I hope we can talk about the next few new models from Google, too. I'm glad the models were finally released because I think it was possible that the models would cover at least a longer period. Not on, there, week, two on which this whole story happened. But, you know, we talked about some kind of, uh, Free Fable website called, yeah, like free Fable. And it was just that the experts wrote that China wasn't that far away from us, yes. And if we stop for a while, well, we mean, the authors are like a collective West that stands up against China. Accordingly, if they stop using advanced models to improve systems, there, cybersecurity and other things, China will soon catch up and will be in no way defended, yes. That we can't close these models for long. Here. And it really happened that the model was released. OpenAI also said in its release that they were being prepared for such a pipeline. They opened it for the State, then they'll open it for everyone in a while. But the long-playing goal is to make their models available, yes. And they want it to be a more understandable story and, eventually, to bring new models to the end-user. That's certainly a pleasure. Although I confess, and the current state of the models we use, even without the Fable, I think it's very and very worthy. Even if we're stuck at this stage for a while, like it's not bad. If we move on, it's the power of myphos Fable. Let's see what GPT5 has six.
Look, to be honest, I have a feeling that this is the Fable situation now, it has created a temporary design of this perception. So, it's four eight Opus, like, or GPT5 five Codex, it's so cool. People in them have a certain time. Maybe it's good that there wasn't some super fast updates of the endless. A huge number of people knew systems were already cool, five Pros were cool, so much as a pep. Not much, she doesn't know how to work. Well, I think she's being cranked inside, and I think five Highs are crambling. What I saw, I think they're cranking, coughing, and there's Codex clicking, and Claude is inside. And, in fact, Fable or five-six, he gives the feeling that the artificial intelligence is already incredibly serious about creating something, incredibly serious. So we went out to that site and we'll see. There is no discussion about what AGI is doing. No more, no discussion of what ISA is doing. I'm a little impressed by this, by the way, because this is the AGI and ISA distraction, it's been undone from all the holes, from all the famous people. Everyone's gone. You sit, you think, wait, if you're all so famous and famous, you know, you can plan, and it's worth billions or trillions of dollars, you know, basically, you don't run the plan. Well, I mean, it's personal to conclude by myself. They just got a little less talkin' about it because it's not clear what to say and it's not clear what to call it at all.
And perhaps we'll have some new words or new ones, perhaps some names in front. Yeah, and I'm personally waiting for the next one, the next such a wave, it's gonna be very interesting. If it comes, it'll start to show up physically artificial intelligence. Everything about some equipment of interesting, cool, well, really cool. There was a leak at Apple's, that they'd have cellphones there, that they'd have some glasses there. I think Apple can't let it go so much, because they're supposed to have some kind of tidy-ass iPad holding cell that's looking around and on. I think it's some kind of a joke, I guess, not more than that. Yeah, I turned Apple Intelligence into iPhone, by the way. I thought I needed him if I never used it in my life, didn't I? And the button is the side of the box, I call it to start the photos. Well, just to take a picture, yeah, inside. And, well, I understand you're not using anything either. And actually, yeah. And it turns out that there's a physical AI when they start showing up, someone's gonna release some kind of working serious devices. And we're thinking about you, and a year and a half ago, I bought this one, well, there, one of the first-time Limitless devices to carry, yeah, which was supposed to be writing, that means, that's all the transcribations, doing all the stuff, doing it. details. That's a kind of culin, right? I mean, there's a corn, this calon has twenty-four for seven, all of it. What's interesting is that when this penis was released, I didn't know what the use scenario would be with him. He doesn't write audio. I mean, I don't get access to audio. He's been in the middle of my head and he knows where I speak and diagnose in my text. I mean, he's transcritibating everything to me. He's showing me the text, and it says in that text that you said it. And there are other Speakers. Speaker one, speaker two, there, speaker three, four, five, fifteen and so on. And by collecting information for two weeks, paying for the unlim, because I was there seven minutes because I was talking all day and, well, at least ten hours a day and so much, Talks, talk. I'm very much of a content-- talking questions- answers to people, and I'm very much in the middle of a different debate. I mean, the content is catastrophic. And I saw the next problem.
I saw the next problem. Turns out he doesn't do all my data. And I think it's one of the fundamental problems of systems in general. That there's no simple system for a normal person to keep him up with his data and deal with them. That's how much I've been trying to find these systems, I haven't found it yet. Everyone says they are. But then, once you're dealing with this real thing, so she's really there, so she's really going to be looking for it to really try to get in and find, really study. Turns out I was writing to him what I said yesterday. He answers me sometimes, sometimes he doesn't answer, and he answers great. If I give him, for example, yesterday from 11 to 1:00, and I say, "In this segment, do me an analysis." And that's kind of something. By the way, it would seem that the models have flew so far, so high speeds, so much quality of everything, but there is no indication of the use, the widespread use of such systems and devices.
So, very micro-use in this subject is some kind of micro-, micro-, micro-level. And now, of course, a very important, interesting story. This is Elnar's special release, he's coming out, so, a few days after this release, this sub-cadast. On the question of whether you have the right and what you have the right to, and how to use artificial intelligence in your work, in your professions. And what you can do. Yeah? I think the subject is very important, too. You'll have to watch this graduation, come to him, if you're interested in other episodes, tell him that. Because we are now, of course, in such an era, well, or six months of certain turbulence, because the main market and the substantive discussions, they are still two thousand twenty-fourth years old. People mix up in terms of who does, tell about the uninterested things, and interesting things can get down sometimes. We've stopped understanding you, and how powerful or cool systems are, or something even starts to distract. That's how Sonnet fifth left. I'm here, Elnar, so it says it. Flagman's, so the model, the working horse Sonnet five. And I-- and it says it, and it says it was such a comment. And somewhere, it even catches Opus four points eight. I threw it out. Well, I mean, logic is very simple to me. Why would I even be a permanent programmer or a system developer, why would I use Sonnet if I could use Opus? Well, Elnar will answer why. Says, "Sasha, Well, as if Opus was to use the tokens, you'd have $20,000 a month.
Limits are big, yeah.
Yeah. And Sonnet can, I don't know, spend a thousand dollars token. But I'm going out-- I'm going out of my position for me in my life, and I need efficiency and quality, right? I mean, it's gonna win while that effect wins. By the way, you know, we talked to you this morning. I came back for something about Fable. So? Well, you just said you had Fable ate all the tokens, so yesterday. I come in, I looked at-- my currents Fable, and I see that until the seventh yuuuuuuuuuuuuuuuuuuuuu, there's nothing
left.
And I thought, and I wanted to do something that would cost a few thousand dollars, and it would be closed. And in general, for a few thousand dollars Fable, I'm not good at it. I mean, maybe not much, but I'm gonna chase some stuff and I'm gonna do it in a little bit. Well, it's like choosing. I mean, like I did yesterday when Fable opened up, I've already not moved my whole Fable task, as I did two--three weeks ago before the blocking. I've moved everything there. Here. I took Fable, so the specific project was a separate, infrastructural study, and I said, "If you're the only one in the infrastructure, you're gonna take care of it." And all the other tasks I've been in, there, Code five High and Opus. Four points eight, uh, Claude.
Surely, Sas, maybe the anthropic will be giving up this trend now, that some model in the subscription will be, there will be no model in the subscription. That's how they say that the seventh July is like getting off the Fable subscription, it's only gonna be on API. And in fact, he's gonna need a lot of money for him. I hope that the rest of the companies will try to keep a large number of models on their subscriptions and thus re-smart users, not all that simple models in the subscription are. You want something good, you know, pay the money on API for the tokens and so on.
Yeah, that's gonna be a new move, of course. Our audience, write, means you're in this part, and you're gonna have to tell us what topics you're interested in. Well, some big, big, big, interesting. We'll have to take them off, mention them. Well, we're waiting for a special release from Tanya, how she's going to be on the agents and set up her, uh, normalizing her entire job. I insist that this should be done without a course.
We'll find out if I can get a course or not.
Bye, everybody.
All of you.