Elon’s xAI STUNS With Grok 4 ‘Smartest AI In The World’ (full replay)

0:00 In a world where knowledge shapes destiny, one creation dares to redefine the future. From the minds at XAI, prepare for Gro 4. This summer, the next generation arrives faster, smarter, bolder. It sees beyond the horizon, answers the unasked, and challenges the impossible. Gro 4. Unleash the truth. Coming this summer. All right, welcome to the Gro 4 release here. Um, this is uh the smartest AI in the world and we're going to show you exactly how and why. Um, and uh, it really is remarkable to see the advancement of artificial intelligence, how quickly it is uh, evolving. Um, I I sometimes think compare it to the growth of a human and how fast a human learns

0:48 and gains conscious awareness and understanding and AI is advancing just vastly faster than any human. Um, I mean, we're going to take you through a bunch of benchmarks that um that Grock 4 is able to uh achieve incredible numbers on. Um, but it's it's it's actually worth noting that like Grock 4 if if given like the SAT would get perfect SATs every time, even if it's never seen the uh the questions before. Um and if even going beyond that to say like uh graduate student exams like the GRE uh it will get near-perfect results in in every discipline of uh of education. So from the humanities to like languages, math, physics, engineering, pick

1:35 anything and we're talking about questions that it's never seen before. These are not not on the internet and it's Grock for is smarter than almost all graduate students uh in all disciplines simultaneously like it's actually just important to appreciate the like that's uh really something um and uh the the reasoning capabilities of Grock are incredible. Um so there's some people out there who who think AI can't reason and look it it can reason at superhuman level. Um so yeah and frankly it it only gets better from here. So we'll we'll take you through uh the Gro 4 release and uh yeah um show you like the pace of pace of progress here. Um like I guess the first part is like in

2:20 terms of the training um we're going from Grock 2 to Grock 3 uh to Grock 4. We've essentially increased the training by an order of magnitude in each case. So it's you know 100 times more training than than Grock 2. Um and uh and that that's only going to increase. Um so it's uh yeah frankly I mean I don't know in some ways a little terrifying but uh the growth of intelligence here is is remarkable. It's important to realize there are two types of training compute. One is the pre-training compute that's from GR 2 to GR 3. Um but for from GR 3 to GU 4 we're actually putting a lot of compute in reasoning in RL. Yeah. And uh just like you said this is literally the fastest moving field and Grock 2 is like the high school student by today's standard. If you look back in

3:07 the last 12 months Grock 2 was only a concept. We didn't even have GR 2 12 months ago. Um and then by training gro that was the first time we scale up like the pre-training we realized that if you actually do the data oblation really carefully and infra and also the algorithm we can actually push the pre-training quite a lot by amount of 10x to make the model the best pre-trained based model and that's why we built Colossus the world's supercomput with 100,000 H100 and then with the best pre-trained model and we realized if you can collect these verifiable outcome reward you can actually train these model to start thinking from the first principle start to reason correct his own mistakes and that's where the gra reasoning comes from and today we ask the question what happens if you take the expansion of classes with all 200,000 GPUs put all

3:52 these into RL 10x more compute than any of the models out there on reinforcement learning unprecedented scale what's going to happen so this is a story of Gro 4 and uh you know Tony uh share some insight with the audience yeah um so yeah let's just talk about how smart off for is so I guess um we can start discussing this benchmark called humanities last exam and this this benchmark is a very very challenging benchmark every single problem is curated by subject matter experts um it's in total 2500 problems and it consists of many different subjects mathematics natural sciences uh engineering and also all humanity subjects so um essentially when when it

4:38 was first released actually like earlier this here uh most of the models out there can only get singledigit accuracy on this benchmark. Yeah. So we can we can look at some of those examples. um you know uh so there there is this mathematical problem which is about natural transformations in category theory and there's this organic chemistry problem that talks about uh electro cyclic reactions and also there's this linguistic problem that tries to ask you about distinguishing between closed and open syllabus uh from a hub Hebrew source text. So you can see also it's a very wide range of problems and every single problem is PhD or even advanced research level problems. Yeah. I mean the these there are no

5:25 humans that can actually answer these can get a good score. I mean if you actually say like any given human um what like what's the best that any human could score? I mean I'd say maybe 5% optimistically. Um so this this is much harder than than what any any human can do. It's it's incredibly difficult and you can see from the the types of questions like you might be incredible in linguistics or mathematics or chemistry or physics or any one of a number of subjects but you're not going to be um at a postgrad level in everything and Gro is a postgrad level in everything like it's it's just some of these things are just worth repeating like Grock 4 is postgraduate like PhD level in everything better than Ph like most PhDs would fail so it's better that

6:10 said I mean at least With respect to academic questions, it I want to just emphasize this point. With respect to academic questions, Grock is better than PhD level in every subject. No exceptions. Um, now this doesn't mean that it's it, you know, at times it may lack common sense and it has not yet invented new technologies or discovered new physics, but that is just a matter of time. um if it I I I think it may discover new technologies uh as soon as later this year. Um and I I would be shocked if it has not done so next year. So I would expect Grock to yeah literally discover new new technologies that are actually useful no later than next year and maybe enter this year. Um and it might discover new physics next year and within two years I'd say almost

6:57 certainly like so just let that sink in. Yeah. So yeah. Um how Okay. So I guess we can talk about the the what what's behind the scene of graph 4. As Jimmy mentioned uh we actually throwing a lot of compute into this training. You know when it started it's only also a single digit sorry um the previous slide sorry. Yeah, it's only a singledigit u number but as you start putting in more and more training compute it started to gradually become smarter and smarter and eventually solved a quarter of the h problems and this is without any tools. The next thing we did was to adding a tools capability to the model. And unlike growth 3, I think you know G3

7:44 actually is able to use C as well. But here we actually make it more native in a sense that we put the tools into training. Uh G3 was only relying on generalization. Here we actually put the tools into training. And it turns out this significantly improves the model's capability of using those tools. Yeah, I remember we had like a deep search back in the days. Yeah. So how is this different? Yeah. Yeah. Exactly. So deep search was exactly the graph three reasoning model uh but without any specific training but we only asked it to use those tools. So compared to this it was much weaker in terms of its tool capability and unreliable and unreliable. Yes. Yes. And and to be clear, like these are still, I'd say, fairly this is still fairly primitive tool use. If you compare it to say the

8:30 tools that are used at Tesla or SpaceX, uh where you're using um you know, finite element analysis and computational flow dynamics and and you're you're able to run uh or say like Tesla does like crash simulations where the simulations are so close to reality that if the test doesn't match the simulation, you assume that the test article is wrong. That's how good the simulations are. So, Grock is not currently using uh any of the tools that the really powerful tools that a company would use, but but that is something that we will provide it with later this year. So, we'll have the tools that that a company has um and and have very accurate physics simulator. Um ultimately, the the thing that will make the biggest difference is being able to interact with the real world via humanoid robots. So you combine sort of

9:16 Grock with with Optimus and it can actually interact with the real world and figure out if if it's hypo if it has if it's formulate an hypothesis and then confirm if that hypothesis is is true or not. Um so we're really you know think about like where we are today. We're we're at the beginning of an immense intelligence explosion. We're in we're in the intelligence big bang right now. Um and the mo we're at the most interesting time to be alive of any time in history. Yeah. Now that said, we need to make sure that the AI is um a good AI. Uh good Grock. Um and the the thing that I think is most important for AI safety, at least my biological neural net tells me the most important thing for AI is to be maximally truth seeeking. So this is this is a very

10:03 fundamental um like you can think of AI as this super genius child that ultimately will outsmart you but you can still in you can instill the right values um and encourage it to be sort of you know uh truthful uh I don't know honorable you know good good things like uh the values you want to instill in in a child that that that grow would grow ultimately grow up to be incredibly powerful um yeah mhm So yeah, so these this is really I say we say we say tools these are say still primitive tools not the kind of tools that um that uh serious commercial companies use but we will provide it with those tools and uh I think it will be able to solve with those tools real world technology problems in fact I'm

10:49 I'm certain of it. It's just a question of how long it takes. Yes. Yes. Exactly. Um so is it just compute all you need 20 right? Is it just compute all you need at this point? Well, you you need compute plus plus uh the right tools and um and then ultimately to be able to interact with physical world. Yes. Um and then I mean we'll effectively have an economy that is well ultimately an economy that is thousands of times bigger than our current economy or maybe millions of times. I mean if you if you think of civilization as percentage completion of the kadeshv scale where kadeshv one is using all the energy output of a planet and kv 2 is using all the energy output of a sun and three is all the energy output of a galaxy. We're

11:35 we're only in my opinion probably like close closer to 1% of kadesev 1 um than we are to 10%. So like maybe a point one one or two% of kadesev one. So um we we will get to um most of the way like 80 90% kadeshv 1 and then hopefully if civilization doesn't self annihilate and then kadesv 2 like it's the the actual notion of of a human economy assuming civilization continues to progress will seem very quaint um in in retrospect um it will it will seem like uh sort of cavemen throwing sticks into a fire uh level of economy um compared to what the future will hold. Um, I mean, it's very exciting. I mean, I I've been at times

12:22 kind of worried about like, well, you know, is this this seems like it's it's somewhat unnerving to have intelligence created that is far greater than our own. Um, and will this be bad or good for humanity? Um, it's like I I I think it'll be good. Most likely it'll be good. Um, yeah. Yeah. But I somewhat reconciled myself to the fact that even if I even even if it wasn't going to be good, I'd at least like to be alive to see it happen. So yeah. So actually one one yeah I think one one technical problem that we still need to solve besides just compute is how do we unblock the data uh data um bottleneck. uh because um when we try to scale up

13:08 the RL uh in this case we did invent a lot of new techniques innovations to allow us to figure out how to find a lot of a lot of challenging RL problems to work on. It's not just a problem itself needs to be challenging but also it needs to be um you also need to have like a reliable uh signal to tell the model you did it wrong, you did it right. This is the sort of the principle of reinforcement learning and as the models get smarter and smarter the number of cool problem or challenging problems will be less and less. Yeah. So it's going to be a new type of challenge that we need to surpass besides just compute. Yeah. Yeah. And we actually are running out of of of actual test questions to ask. Uh so there's like even ridiculously

13:54 questions that are ridiculously hard if not essentially impossible for humans that are written down questions. um are uh becoming swiftly becoming trivial for for AI. Um so then there's um but you know the the one thing that is an excellent judge of things is reality. So because physics is the law ultimately everything else is recommendation. You can't break physics. Um so the ultimate test I think for whether an AI is um the the ultimate reasoning test is reality. So you invent a new technology like say improve the design of a car or a rocket or um create a new medication uh that and and and does it work? Um does does the rocket get to orbit? Does the does the car drive? Does the medicine work?

14:40 Whatever the case may be. Um reality is the ultimate judge here. Um so it's it's going to be reinforcement learning closing the loop around reality. We asked the question how do we even go further? So um actually we are thinking about now with single agent we're able to solve 40% of a problem. What if we have multiple agents running at the same time? So this is what's called test and compute and as we scale up the test and compute actually we are able to solve almost more than 50% of the uh text only subset of the h problems. So it's a remarkable achievement I think. Yeah. Yeah. This is this is this is insanely difficult. the these are it's it's so what we're saying is like a majority of the of the of the textbased um of humanities you know scarily named

15:25 humanities last exam um GR for can solve and you and you can try it out for yourself um and the with the GR 4 heavy what what it does is it spawns multiple agents in parallel and uh all of those agents do do work independently and then they compare their work and they they decide which one like it's like a study group um and it's not as simple because majority vote because uh often only one of the agents actually figures out the trick or figures out the solution um and and and but once they share the the trick or or or figure out what what the real nature of the problem is they share that uh solution with the other agents and then they compare they essentially compare notes and then and and then yield yield an answer. So that's that's the the heavy part of Grock 4 is is

16:11 where we you scale up the test time compute by roughly an order of magnitude uh have multiple agents uh tackle the task and then they compare their work and they they put forward what they think is the best result. Yeah. So we will introduce graph 4 and graph for heavy. Sorry can you click the next slide? Yeah. Yes. So yeah, so basically graph 4 is a single version a single agent version and graph 4 heavy is the multi- aent. So let's take a look how they actually do on those exam problems and also some real real life problems. Yeah. So we're going to start out here and we're actually going to look at one of those HLE problems. This is uh actually one of the easier math ones. Uh I don't really understand it very well. I'm not that smart. But I can launch

16:57 this job here and we can actually see how it's going to go through and start to think about this problem. Uh while we're doing that, I also want to show a little bit more about like what this model can do and launch a uh Gro 4 heavy as well. So everyone knows Poly Market. Um it's extremely interesting. It's the you know seeker of truth. It aligns with what reality is most of the time. And with Grock, what we're actually looking at is being able to see how we can try to take these markets and see if we can predict the the future as well. So, as we're letting this run, we'll see how uh Gro for Heavy goes about uh predicting the, you know, the World Series odds for like the current teams in the MLB. And while we're waiting for these to process, we're going to pass it over to Eric and he's going to show you an example of his.

17:43 Yeah. So uh I guess one of the coolest things about uh Gro 4 is its ability to understand the world and to solve hard problems by leveraging tools like Tony discussed. And I think one kind of cool example of this we asked it to generate a visualization of two black holes colliding. Um and of course you know it took some there are some liberties. It's in my case actually pretty clear in its thinking trace about what these liberties are. Uh, for example, in order for it to actually be visible, you need to really exaggerate the the scale of the, you know, the uh the uh the waves. And yeah, so here's like, you know, this kind of in action. Um, it exaggerates

18:29 the scale in like multiple ways. uh it drops off a bit less in terms of amplitude um over distance and uh but yeah we can kind of see uh the basic effects that you know are actually like you know correct. It starts with the inspiral it merges uh and then you have um the ring down and like this is basically um largely correct um yeah uh modulo some of the simplifications that need to do um you know it it's actually quite explicit about this you know it uses like post postneonian approximations instead of actually like computing the general relativistic effects like uh near the center of the black hole which is you know incorrect um and you know will lead to you know

19:15 some incorrect results but the overall you know visualization is uh yeah it's basically there um and you can actually look at the kinds of resources that it references. So here it um it actually you know it obviously uses search it gathers results from a bunch of links but also reads through a undergraduate text in analytical analytic gravitational wave models. Uh it um yeah it um reasons quite a bit about uh uh the actual constants that it should use for a realistic simulation. Uh it references uh I guess existing real world data. Um and yeah I it yeah it's a it's a pretty good model. Uh yeah,

20:01 but like actually going forward we can we can we can plug we can give it the same model that physicists use. So it it can run the the same level of compute that uh so leading physics researchers are using and and give you a physics accurate black hole simulation. Exactly. Just right now is running in your browser. So yeah, this is just running in your browser. Exactly. Pretty simple. So swapping back real quick here, we can actually take a look. The math problem is finished. uh the model was able to uh let's look at its thinking trace here so you can see how it went through the problem. I'll be honest with you guys, I really don't quite fully understand the math, but what I do know is that I looked at the answer ahead of time. Um and it did come to the the correct answer here in the final part here. Um we can also come in and actually take a

20:46 look here at our uh our World Series prediction. Um and it's still thinking through on this one, but we can actually try some other stuff as well. So we can actually like try some of the X integrations that we did. So we worked very heavily on working with uh all of our X tools and building out a really great X experience. So we can actually ask you know the model you know find me the XAI employee that has the weirdest profile photo. Um so that's going to go off and start that and then we can actually try out you know let's uh create a timeline based on xost uh detailing the you know changes in the scores over time and we can see you know all the conversation that was taking place at that time as well. So we can see who were the you know announcing scores and like what was the reactions at those times as well. Um so we'll let that go through here and process. And if

21:34 we go back to uh uh this was the uh Greg Yang photo here. So, uh, if we scroll through here, whoops. So, Greg Yang, of course, who has his favorite, uh, photograph that he has on his account. Um, that's actually not how he looks like in real life, by the way. Um, just so aware, but is quite funny. But it had to understand that question. Yeah. Which is the that's the wild part. It's like it understands what is a weird photo, what is a weird photo? Yeah. What is a less or more weird photo? It goes through, it has to find all the team members, has to figure out who we all are. it, you know, searches without access to the internal XAI personnel logs. It's literally looking at the just at the internet. Exactly. So, you could say like the weirdest of any company. Yeah. To be clear. Exactly. And uh we can also take a look

22:19 here at the uh question here for the humanities last exam. So, it is still researching all of the historical scores. Um but it will have that final answer here soon. But we can while it's finishing up, we can take a look at one of the ones that we set up here a second ago. And we can see like you know it defines the date that like Dan Hendricks initially announced it. We can go through we can see you know uh OpenAI announcing their score back in uh February. And we can see you know as progress happens with like Gemini. We can see like Kimmy uh and we can also even see you know the leaked benchmarks of uh what people are saying is you know if it's right it's going to be pretty impressive. So pretty cool. Um so yeah I'm looking forward to seeing how everybody uses these tools and gets the most value out of them. Um but uh yeah, it's been great. Yeah. And we're gonna close the loop

23:05 around usefulness as well. So it's like it's not just books smart but actually practically smart. Exactly. And we can go back to the uh the slides here. Yeah. Cool. Um so the so we actually evaluate also on the multimodel subset. So on the full set uh this is the number on the h exam. Uh it you can see there's a little dip on the numbers. uh this is actually something we're improving on which is the multimodel uh understanding capabilities but I do believe uh in a very short time we're able to really improve and got much higher numbers on this um even higher numbers on this benchmark. Yeah, this is the we we still like what what is the biggest weakness of Grock currently is that it's it's sort of partially blind. it can't it's it's

23:52 image understanding um obviously in its image generation uh need to be a lot better um and that um that that's actually um being trained right now. So um GR 4 is based on version six of our foundation model um and we are training version 7 uh which we'll complete in a few weeks um and uh that that'll address the weakness on the vision side. Mhm. Just to show off this last here. So the the prediction market finished uh here with a heavy and we can see uh here we can see all the tools and the process it used to actually uh go through and find the right answer. So it browsed a lot of odds sites. It calculated its own odds comparing to the market the market to find its own alpha and edge. It walks you through the entire process here and

24:38 it calculates the odds of the winner being like the the Dodgers uh and it gives them a 21.6% 6% chance of uh winning uh this year. So, and it took appro approximately four and a half compute. Yeah, that's a lot of thinking. Yeah, we can also look at all uh the other benchmarks besides HRE. Um as it turned out, Graph 4 excelled on all the reasoning benchmarks that people usually test on um including GBQA which is a PhD level uh problem sets. Uh that's easier compared to HRE. Um on Amy 25 America invitation mathematics exam we with graph for heavy we actually got a perfect score. Um also on some of the coding benchmark called live coding bunch um and also on uh hmmt hover math

25:28 uh MIT exam and also USMO uh you can see actually um for on all of those benchmarks we often have a very large leap against the second best uh model out there. Yeah, it's I mean really we're going to get to the point where uh it's going to get every answer right in every exam. Um and where it doesn't get an answer right, it's going to tell you what's wrong with the question or if the question is ambiguous, disambiguate the question into answers A, B, and C and tell you what it would what answers A, B, and C would be with a disambiguated question. Um so the only real test then will be reality. Uh can I make useful technologies, discover um new science? that'll actually be the the only thing left because human tests will simply not be um meaningful.

26:13 Yeah, we need to make an update to HR very soon given the current rate of progress. So yeah, it's super cool to see like multiple agents that collaborate with each other solving really challenging problems. Uh so we're going try this model. Uh so it turned out it's available right now. Uh if we advance to the next slide uh where there is a super Grock heavy tiers that we're introducing where you're able to access to both Grock 4 and Grock 4 heavy uh where you're actually going to be the taskmaster of bunch of little Grock research agent to help you you know become smarter do all the little research and save hours of times of uh you know going through mundane tasks uh and it's available right now uh so yeah so u we we did limit usage during the the demo so didn't it didn't break

26:58 the demo because all these this all this stuff is happening live. So it's not there's not nothing canned about the any of the tests that we're doing. Um so uh after the after the uh demo is done we'll we'll allow we'll enable more subscribers for CGOR. So if you can't subscribe right now just try in half an hour. It should work. Um so uh and now let's let's get into voice. Great. So many of you have been enjoying our voice mode and we've been working hard to improve the experience over the past couple months. Um, we have cut latency in half to make it much snappier. And today we're excited to announce a set of new voices that have exceptional naturalenness and procity. You might have noticed the movie trailer voice that opened up the live stream. That is one of our new voices, S, who's got that

27:45 epically deep tone. And we're also excited to introduce Eve, a beautiful British voice who's capable of rich emotion. Aman, would you like to introduce Eve? Absolutely. Yeah. Let's get into the demo. Hey, Eve. Millions of people are watching us right now. Are you excited? Oh, I'm thrilled. Really? It's like being on stage at the Old Vic with a sea of faces out there. Just imagine all those eyes and ears tuned in. It's a proper buzz, isn't it? What do you reckon we should chat about to keep this lot entertained? You know, we have a lot to talk about, but I'll be honest. I'm feeling a little nervous. Can you whisper something soothing to calm me down? You've got this. Just having a quiet chat like we're tucked away. The world's not out there. Feel that calm. Let's

28:30 keep nice. What's on your mind? You know what? I'm actually feeling quite relaxed to try this um something crazy. So, can you give us an opera on Diet Coke? Oh, an opera on Diet Coke. That's a riot. Picture this. A shimmering silver cannon glistening under stage light. Oh, Diet Coke, thou elixir divine with bubbles that dance in a sparkling line. Thy crisp cool kiss on lips so fine. Oh, aspartame and bro forever mine. How's that for a mad little area? Want me to belt out another verse or switch up the tune? Okay, let's try switching it up and just start singing. Right then, let's have a go. Oh, silver

29:17 can, you gleam in the light. Diet Coke, my heart's delight. Okay, I won't touch it. It's very, very poetic. Um, yeah, thank you. Feeling quite nice now. Cool. Um, so that's Eve. And, um, so something else we do here is is a little test each day to check on the natural flow of our conversations against some of our competitors. So, we figured why not try that live on the stream today. Should I start? Hey, I'm going to say a number. And can you say the same number back to me? One. One. Got it. This is like a fun little game of echo. What's next? Two. So, you said two. Three. You did. Three. Four. Nice and steady. What's next? Um. Uh. Five. Hey. Nailed it. This is pretty fun. Feel free to keep going or switch it up.

30:03 Hey, Grock. Hey, Grock. Hello there. How can I help you today? We're going to do a little test. Can you repeat after me? One. One. All right. What's next? Two. Two. What's on your mind? Three. Three. Need anything else? Four. Four. How can I assist you? Five. Five. What's next? So, as you can see, um, Groc was, uh, snappier. It didn't interrupt me. And the proced choices. I think we shooting for something more calm, smooth, more natural versus something that's more poppy or artificial. So, we'll keep improving. All right. Thanks, guys. Yep. So, since the launch of the voice model, uh, we actually see the 2x faster end to end latency. uh in the last 80 weeks five different voices and also 10xed the active user so Grock voice is taking off now um if you think about

30:51 releasing the models this time we're also releasing Grock 4 through the API at the same time so if we go to the next two slides so uh you know we're very excited about you know what all developers out there is going to build so you know if I think about myself as a developer what the first thing I'm going to do when I actually have access to the graph for API benchmarks so we actually ask around on Xplatform what is the most challenging benchmarks out there that you know is considered the holy grail for all AGI models. Uh so turn out AJI is in the name RKGI. So uh the last 12 hours uh you know kudos to Greg over here in the audience. Uh so who answered our call take a preview of the Gro for API and independently verified you know the Gro for's performance. So initially

31:38 we thought hey gravoy is just you know we think it's pretty good it's pretty smart uh it's our nextgen reasoning model spend 10x more compute it can use all the tools right but turned out when we actually verify on the private subset of the rkgi v2 it was like the only model in the last three months that breaks the 10% barrier and in fact was so good that actually get to 16% well 15.8% 8% uh accuracy 2x of the second place that is the cloud for opus model um and it's not just about performance right when you think about intelligence having the API model drives your automation it's also the intelligence per dollar right if you look at the plots over here the gro is just four just in the league of its own um all right so enough of benchmarks uh over

32:24 here right so what can gro do actually uh in the real world so uh we actually uh you know contacted the folks from uh Endon Labs uh who you know were you know gracious enough to you know try to grow in the real world to run a business. Yeah, thanks for having us. So I'm Axel from and Labs and I'm Lucas and we tested Grog 4 on vending bench. Vending bench is an AI simulation of a business scenario uh where we thought what is the most simple business an AI could possibly run and we thought vending machines. Uh so in this scenario the the Grock and other models need to do stuff like uh manage inventory, contract, contact suppliers, set prices, all of these things are super easy and all of they like all the

33:10 models can do them one by one. But when you do them over very long horizons, most models struggle. Uh but we have a leaderboard and there's a new number one. Yeah. So we got early access to the GR 4 API. Uh we ran it on the vending bench and we saw some really impressive results. So it ranks definitely at the number one spot. It's even double the net worth which is the measure that we have on this value. So it's not about a percentage on a uh or a score you get but it's more the dollar value in net worth that you generate. So we were impressed by Grock. It was able to formulate a strategy and adhere to that strategy over long period of time much longer than other models that we have tested other frontier models. So it managed to run the uh simulation for double the time and score yeah double the net worth and it was also really

33:56 consistent uh across these runs which is something that's really important when you want to use this in the real world and I think as we give more and more power to AI systems in the real world it's important that we test them in scenarios that either mimic the real world or are in the real world itself. Um because otherwise we we fly blind into some uh some things that uh that might not be great. Yeah, it's uh it's great to see that we've now got a way to pay for all those GPUs. So, we just uh need a million vending machines. Definitely. Um and uh we could make uh $4.7 billion a year with a million vending machines. 100%. Let's go. They're going to be epic vending machines. Yes. Yes. All right. We are actually going to install vending machines here. Uh like a lot of them. We're happy to supply them. All right. Thank you. All right. I'm looking forward to seeing

34:42 what amazing things are in this vending machine. That's that's for uh for you to decide. All right. tell the AI. Okay, sounds good. Um, all right. Yeah, I mean, so we can see like Grock is able to become like the co-pilot of the business unit. So, what else can Grock do? So, we're actually releasing this Grog if you want to try uh right now to evaluate run the same benchmark as us. Uh, it's on the API um has 256k contact length. So, we already actually see some of the early early adopters to try Guac for API. So uh our Palatoto neighbor ARC Institute which is a leading uh biomedical research uh center is already using seeing like how can they automate their research flows with gro. Uh it turned out it performs is able to help the scientists to sniff

35:28 through you know millions of experiments logs and then you know just like pick the best hypothesis within a split of seconds. uh we see this is being used for their like the crisper uh research and also uh you know block four independently evaluated scores as the best model to exam the chess x-ray uh who would know um and uh uh on in the financial sector we also see you know the graph with access to all the tools real-time information is actually one of the most popular AIs out there so uh you know our graph is also going to be available on the hyperscalers so the XAI enterprise sector is only, you know, started two months ago and we're open for business. Um, yeah. So, um, the other thing, uh, we talked a lot about, you know, having Grock to make games,

36:14 uh, video games. Uh, so Denny is actually a, uh, video game designers on X. So, uh, you know, we mentioned, hey, who want to try out some, uh, uh, Grock for uh, preview APIs, uh, to make games? And Danny answered the call. Uh, so this was actually just made first-person shooting game in a span of four hours. Uh so uh some of the actually the unappreciated hardest problem of making video games is not necessarily encoding the core logic of the game but actually go out source all the assets all the textures files and and uh you know to create a visually appealing game. So one of the core aspect guac for does really well with all the tools out there is actually able to automate these like asset sourcing capabilities. So the

36:59 developers you can just focus on the core development itself rather than like you know so now you can run a you know entire game steals with game of one but with like one person and then uh you can have gro 4 to go out and source all those assets do all the maintaining task for you. Yeah. The now the next step obviously is for gro to uh play be able to play the games. So it has to have very good video understanding so it can play the games and interact with the games and actually assess what whether a game is fun and and and actually have good judgment for whether a game is fun or not. Um so with the with version seven of our foundation model which finishes training this month and then we'll go through post training RL and whatnot. Um that that will have excellent video understanding. Um, and with the with the video understanding

37:45 and the and improved tool use, for example, for video games, you'd want to use, you know, Unreal Engine or Unity or one of the one of the the main graphics engines. Um, and then gen generate the generate the art, uh, apply it to a 3D model, uh, and then create an executable that someone can run on a PC or or a console or or a phone. Um, like we we expect that to happen probably this year. Um, and if not this year, certainly next year. U, so that's uh it's going to be wild. I would expect the first really good AI video game to be next year. Um, and probably the first uh half hour of watchable TV this year and probably the first watchable AI movie next year. Like things are really

38:30 moving at an incredible pace. Yeah. When Gro is 10xing world economy with vending machines, it would just create video games for human. Yeah. Yeah, I mean it went from not being able to do any of this uh really even 6 months ago to to what you're seeing before you here and and from from very primitive a year ago uh to making a a sort of a 3D video game with with a few hours of prompting. Yep. I mean yeah just to recap so in today's live stream we introduced the most powerful most intelligent AI models out there that can actually reason from the first principle using all the tools do all the research go on the journey for 10 minutes come back with the the most correct answer for you. Um so it's kind of crazy to think about just like four months ago we had Gro 3 and now we

39:16 already have Gro 4 and we're going to continue accelerate as a company XAI we're going to be the fastest moving a AGI companies out there. So what's coming next is that we're going to you know continue developing the model that's not just you know intelligent smart think for a really long time spend a lot of compute but having a model that actually both fast and smart is going to be the core focus right so if you think about what are the applications out there that can really benefit from all those very intelligent fast and smart models and coding is actually one of them yeah so the team is currently working very heavily on coding models um I think Uh right now the main focus is we actually trained recently a specialized coding model which is going to be both fast and smart. Um and I believe we can

40:03 share with that model with you guys with all of you uh in a few weeks. Yeah. Yeah. That's very exciting. And uh you know the second after coding is we all see the weakness of Gro 4 is uh the multimodal capability. So in fact uh it was so bad that you know Grock effectively just like looking at the world squinting through the glass and like see all the blurry uh you know features and trying to make sense of it. Uh the most immediate improvement we're going to see with the next generation pre model is that we're going to see a step function improvement on the model's capability in terms of the image understanding video understanding and audios. Right? It's now the model is able to hear and see the world just like any of you. Right? And now with all the tools at this command with all the other agents it can talk to. Uh you know so

40:50 we're going to see a huge unlock for many different application layers. uh after the multimodal agents what's going to come after is the video generation and we believe that you know at the end of the day it should just be you know pixel in pixel out u and um you know imagine a world where you have this infinite scroll of content in inventory on the X platform um where not only you can actually watch these generate videos but able to intervene create your own adventures the future is going to be wild um and we expect to be training video model with uh over 100,000 GB200s uh and uh to begin that training within the next three or four weeks. So, we're we're confident it's going to be pretty spectacular in video generation and

41:36 video. So, let's see. So, that's uh anything you guys want. Other than that, I guess that's it. Yeah, it's it's a good model, sir. It's a good Well, we're very excited for you guys to try Gro 4. All right. Thanks everyone.