Limitless Podcast - The Point of No Return: GLM 5.2 Approaches the Frontier

Episode Date: June 23, 2026

The Chinese open-weight AI model GLM 5.2 compares with leading models from OpenAI and Anthropic on coding and development tasks. Today, we cover the shift toward cheaper AI models, the difference between open weights and closed models, and the U.S. government’s reported ban on Anthropic’s most powerful model after security testing.------🌌 LIMITLESS HQ ⬇️NEWSLETTER:    https://limitlessft.substack.com/FOLLOW ON X:   https://x.com/LimitlessFTSPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQAPPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890RSS FEED:           https://limitlessft.substack.com/------TIMESTAMPS0:00 China’s Open Source Comeback3:05 Benchmarks and Cost8:15 Markets10:51 A Six-Month Model Gap13:56 The Fable 5 Ban15:56 Public Access and Competition19:26 Open Source vs Open Weights21:09 Multi-Model Routing Arrives25:41 Regulation and the Road Ahead28:31 Closing------RESOURCESJosh: https://x.com/JoshKaleEjaaz: https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠

Last week, a Chinese company released a free AI model that is as good as Anthropics' best model. It also beats ChatGPT-515 at writing and coding, but it comes to the twist. It's a sixth of the price, and it's completely open source. You can download it and run it at home. Now, in that same week, the United States government banned Anthropics' most powerful model, Fable 5, after someone revealed that an unrestricted version of it had hacked into the national security agency's systems. I think we've reached a point of no return. And not to sound dramatic, but in six months, it is very realistic that we will have open source or open weight models that are accessible to anyone in the world with an internet connection and, I don't know, 5 to 10K to run at home, that they can fine tune to do anything.

And it's mythos grade level models. These are the same models that we're hearing rumors and reports from verified that they can exploit some of the most secure systems in the world faster than any other exploiter has been able to. to do in the past. And I think we're going to look back on 2026 as the moment or the year that everything really changed and the point where humanity as itself really needs to focus on safeguards and figuring out how to regulate and release these AI models in the future. So we've reached a convergence of this really interesting trend where the most powerful models of the world are freely available and open source available for anyone to access. And the government, the United States specifically has an off switch for their most powerful model. Yeah, it's been a couple of months, it seems, since we've had some news on the frontier of

China. And you kind of forget about them every couple of weeks where they just kind of disappear, they quiet down. The new models come out. We see the fables. We see the mythos of the world. But then out of nowhere, they strike back. And seemingly every single time, it comes as a surprise at how powerful these new models have become. So to start with this, we have a new model from our favorite company to pronounce, Chi Pooh, the Chinese company. I feel like I wanted it in my dog. that is such a cute name. But GPU is doing something not so cute. They're actually releasing a model named GLM 5.2,

which kind of blew everyone's expectations out of the water. I remember way back like six months ago when Deep Seek was doing this. Like Deep Seek would release the model. Everyone was like, wait, you did what with what? And that's what this model feels like again. We're getting that moment again because this is an open weights model, which is not to be confused with open source. And we'll talk about that in a little bit.

But this is an open weights model that is, if I'm correct about this, within one single point of the SwayBench Pro benchmark, which is the benchmark that a lot of people use for coding. Oh, yeah. Of GPT 5.5, the frontier coding model from OpenAI. And that comes to surprise because the cost, well, one, if you run it locally, is free. But two, if you run it on a server is, like you said earlier, you just one sixth of the cost. So you're getting an incredible amount of coding capability for something that costs a fraction

of what it costs if you were to go to one of these larger language models. And it seems to work almost as good if I'm right. And this comes as a surprise to most people because every time we start to count China out, we're like, no, surely they can't catch up. They continue to chip away at this frontier. There's a few things that people will jump to immediately. Okay, one, that these benchmarks can be easily gamed. We're going to show you a few examples of benchmarks that couldn't be gamed and jLM5.1-2

performs really, really well. But the second thing is the cost. Cost has become a really important point of discussion amongst enterprises specifically that are spending hundreds of millions of dollars per year to access Claude and GPT. It's just too much money for them to spend in terms of the return on investment that they're getting in work that they actually see. So what they're now turning towards is these free open source models, primarily designed and made by Chinese AI labs that can cut costs down drastically.

Just last week, we had Microsoft announced that they're replacing their co-pilot LLM with not chat GPT, with not Claude, but with Deep Seek itself. So the point is, this comes at a very important time where cheaper models are getting a lot of attention. So now when we look at GLM 5.2 specifically, it is five to seven times cheaper the GPT 5.5 and Claude Opus 4.8, but performs, as we're seeing on the benchmarks right here,

almost as good as each of these models, specifically at the metric that is the most important, which is coding. Now, a lot of skeptics quite rightly were like, I don't know if this is actually true. like let me test it against a few other independent benchmarks. It came up pretty high. So if you look at the front end development when it comes to like website design, GLM 5.2 Max is just below Fable 5.

We're not even talking about Opus 4.7 or 4.8 anymore, which it absolutely beat. And then when we're looking at like anecdotes or feedback from like distinguished individuals in the Western frontier, so right now we're looking at a tweet from the CEO of Vassel, he goes, I'm genuinely impressed, almost shocked at how good GLM 5.2 is at coding. So this is feedback from real people using this for real use cases. For the last three years, Josh, we've basically been told that the hundreds of billions of dollars that is being spent on AI CAPEX is for one single reason only to gain a moat ahead of any other model provider. So we spend all this money on compute to train a frontier AI model and that moat,

it doesn't matter what other companies do in China. We will have the best model and that's enough for us. This release from GLM, from Chippoo, with GLM 5.2, basically shows us the opposite. For a fraction of the cost, you can create a near-frontier model that does like, I don't know, 95% of the work. And so it brings into question the valuation between these companies. Should they be spending this amount of money, or can we just do it for a lot cheaper like these Chinese AI labs?

Yeah, well, the large AILabs, I'm not sure they have a choice. I mean, it's just that you have to continue to purchase the frontier forward, whether you like it or not. But I think what we're seeing is a lot of these questions that we were excited. to see play out, we're starting to get answers to. Now, it's less China versus America and more open source versus closed source because, I mean, the open source models are coming from inside too. We have Nvidia. They're working on open source models that are incredible and they're making progress in that front. We have Apple now who has like an actually functional Siri on everyone's

hardware device that runs essentially for free. So they're slowly starting to nibble away at this, I guess the lower bottom of the barrel set of use cases. And then we have China, which is GLM, That's Deep Seek, that's these larger models, where they're actually competing on the frontier. So these big frontier private models are facing heat both from the lower end of the stack, but also right at the top where these benchmarks sit. And we're going to see how that plays out economically. For, in the case of Jipu, at least, it's been playing out pretty well. And we probably should talk about the stock a little bit.

Believe it or not, this company is publicly traded, not here in the United States, but this is publicly traded, at least in China. and it's kind of, what is that, 1,500% on the year. That's like a crazy return and some interesting facts about this return. And it's so funny to see kind of, I guess, how inefficient Chinese markets are. Also, note that the start you're seeing on screen, they have a lunch break in their stock market. I didn't know this. And then labeled it.

Like, I didn't realize that Chinese stock markets had an hour long lunch break in the middle of the day. So that's cute and that's fun. But the numbers are pretty outrageous. when we trade, when we talk about expensive companies, we talk about SpaceX, who's trading, what is it like, a very high multiple towards earnings. And what we have with G-Boo and this company that it's kind of owned by, Knowledge Atlas Technology, is currently trading at about $136 billion market cap. It made $170 million, or $107 million, I should say, in the full year of 2025. That means it trades 1,300 times sales, which is just this unbelievably high multiple on

this company. And I think it's a testament to the, I guess, the lack of availability to get AI exposure in Chinese markets, but also the confidence and the excitement and enthusiasm they have around companies like this. That was just an interesting thing to see. Yeah, I mean, at this valuation, it's about, what is that, like a fifth of Anthropics valuation right now, which is, I think, around a trillion dollars. So again, like, it begs the question, is Chinese AI blabs underpriced or are American companies overpriced? And I'm curious to hear what listeners of the show actually think. I tend to think that they probably need to meet somewhere in the middle. We were actually saying before we started recording, could you imagine the reaction

to this news if Anthropic was a publicly traded company? And a new free-d open source model that was freely accessible to anyone could achieve pretty much 95% of the capability of Opus 4.8. I wonder what that would have done to the stock price in like a fair market value. But crazy to see nonetheless. So if we're looking at a few different metrics that compare cost and performance, just quickly to run you guys through this, for input versus output tokens for a million tokens, you're looking at around $1.50 to $4.50 when it comes to cost.

Now, comparing that to Opus 4.8, that's around, I believe, $5 versus $25. So again, we're achieving that three to five X cheaper when it compares to a model of similar performance and capability. Now, I was skeptical of the benchmarks, and I have a new favorite benchmark to compare it again, which is called DeepSuite. Deep Sway is basically a benchmark that gives no models any answers. Typically with a benchmark, you have an answer sheet, and it can kind of cheat and look at it and figure out a way to get to that answer. There's no answer sheet for this one, so it's a very accurate test of how good your model is at coding. For DeepSwee, GLM 5.2, achieved a very modest fifth place. Now, that is probably, or rather fourth place,

Fifth place, fifth place. And that is a pretty accurate standing of how agenetic coding looks like for this particular model. It's the highest number one place for open source model. It absolutely crushed Kimmy K2 by 17 percentage points, so a very clear lead. And it's great to see how it weighs up. Like, it may not be front-air capability, but if you want a workhorse, if you want an agent that basically works overnight and isn't going to break the bank, GLM 5.2 is probably something that you can look at. Another thing is it's really good at front-end web development. So if you're looking at this screen right now, the website that you're seeing was completely one-shottered in about 10 minutes from this one single model, GLM 5.2. And repeatedly, a cross-design benchmark, arena benchmark was another one that I saw. It performs really highly, in some cases, beating Fable 5.

So it's a really good front-end design model if that is something of interest. Then the final one, because I know a lot of listeners on this show is like, you know, how good are these models like trading, investing, money for you. Well, there's this very famous benchmark, which is called the vending benchmark, which basically allows an AI model to control a theoretical $10,000 and see if it can make money by stocking a vending machine and then conducting sales, managing inventory against competition. It achieved second place right behind Claude Opus 4.7, which is the current leading model. So it's also pretty good at making money as well.

Yeah, and it also has a very clear roadmap to continue to be good and to get even better. There's an interaction actually between Elon Musk and the CEO of Z.AI, who is creating these models. So this guy asked, what's your current timeline for China to reach Fable Class? GLM 5.2 certainly shortened the gap. And then Elon said probably Q1. And then the CEO said won't take that long, which means they expect us to get a new Fable Class level model that's open weight and open source within the next six months, which is incredibly compelling because that is going to be served up as open weights. And as you know, with open weights, you can actually run it on your own hardware. But the question is, do you actually want to run this on your hardware?

I see on Twitter all the time, people who are spending tens of thousands of dollars to get those Max Studios. They're stacking them up in their offices. They're trying really hard to run these models locally. And I hate to break it to you, but the math ain't really math in on this so well. So there's a suite by Max Weinbach, I thought was great. And it says the minimum to run the model is about $20,000 in hardware. And you get about 20 tokens per second out. For $20,000, that's pretty slow.

It's not thinking that fast. And if you have these really long chain of thoughts, these long reasoning traces, it's going to take you a very long time to get an answer. That involves deep thinking. So for about $20,000, you can get close to $35 billion tokens, and that's a 12 to 1 input-to-output ratio,

assuming you have a good token caching setup. So he's saying if you ran the hardware 24-7 with zero downtime, it would take roughly five and a half years just to break even. And that right there, why. Open weights models are incredible, but you're probably better off getting it served directly from their servers from the cloud instead of running your own, because not only do you have to deal with the complexity, you have to power it all on, you have to deal with hardware stuff, and you have to worry about getting the actual hardware, because Lord knows, getting those computers now

is not as easy as it used to be. So interesting note on cost on how available these are and accessible these are in a relative basis. And the Chinese companies themselves are willing to subsidize these costs, just to be clear, like to play around with kids. Kimi K-2.7, which is their frontier model, I've been able to access it and use it since they launched it. And I've been free using it to kind of like do research and all that kind of stuff. And I've never once been charged for it. So there's a high subsidy coming from like the Chinese side of things as well. The other thing I'll say is these numbers may look big, right? Like who on earth is spending $20,000 to get hardware that you can like run at home to run these models open source?

But the idea is six months from now, 12 months from now, these very same models will be distilled enough, so that means it can maintain its intelligence, but good enough to run on your local hardware at home a custom PC or maybe even your laptop. The trend that we're undeniably seeing with these open source models in particular is higher intelligence for lower cost hardware, and if that trend continues, we will end up seeing this model that we're talking about today being able to run off your handset. So it's something that seems unfeasible right now to access, but further on down the line, open source, in my opinion, is pretty undeniable. You'll be able to run it at home, and that's pretty good. But moving on, the reason why we wanted to write this episode

is there's a convergence of two trends, right? So last week, we had a lot of reporting around Fable 5 being bad by the United States government. The primary reason is the United States government does not think the model is safe. If placed in a malicious actor's hands, would be able to be used against government systems, hack, exploits, all that kind of stuff. and it's proven itself on internal testing. And the most recent revealing was a quote from a senator saying that the head of the NSA explained in a red team exercise, which is like a controlled environment, that Claude Mythos 5 was able to breach all of its systems.

And typically it would take months for an individual expert to do that. It did it in hours. And this is just a crazy story and headline to read. They've switched it off. It's not accessible to anyone. If you go on cold right now, you're unable to access Fable 5. But the point is, these two trends have converged at the same time, and it's important to discuss this because very soon in a few months' time,

as that Elon tweet showed, we're going to end up with Mithosgrade level models that are freely available to anyone, subsidized by China or available to run at home for 10K. And that is pretty scary, I guess. Yeah, is that the lead now? Are we at six months? Does that feel about right? Like if they release Mythos class by the end of this year, and then that gives, I guess, an open AI and anthropic a six-month head start to continue to progress power.

And then the head of Chippoo has set it. So, yeah, so it seems like that's about right currently. We're really like a six-month window between us and the current Bleeding Edge open source. I could see that kind of getting closer and closer. It feels like they're right on the tail. Of course, understanding what's going on internally would be very helpful to know because I'm sure GPT 5.5. we know we're getting 5.6 pretty soon. I'm sure Anthropic is working on something even more powerful than mythos.

And it feels like we don't really have a choice but to continue progressing as fast as we are. Otherwise, these are going to catch up. And they won't have the guardrails that are put in place currently by the frontier models. Now, what's happening currently is we're seeing this fork in terms of these private models where only people internally are now able to use them and anyone out in the world is getting, I guess, kind of disabled. They're getting like a handicap because they're not actually able to access these frontier models. So we're seeing this weird crossroads where there's a small subset of people that work internally within Open AI, within Anthropic that are getting access to these models.

The government is limiting their public use, which means the public is getting left behind. And then China's coming up and they're saying, hey, in six months, we're going to be right here at your head. So it's this really interesting dynamic that's at play. And we're going to really have to closely monitor this as these new frontier models continue to be released. Because you have to assume, even though the world isn't using mythos or fable, they're continuing to iterate and to build better models. They're not just going to stop because of this. Same with Open AI, same with all the other frontier labs. So the question is, are these models going to be held privately for just a small subset of people to use?

Or is there going to be this path forward in which the public can use them? I think everyone's hope is that there is a path forward. But currently we're at this weird standstill where it feels like China's kind of breathing down your neck here. Well, the irony also is if the government is just going to come in and switch off the frontier model, it's going to push companies to use open source models. Like, imagine you're an enterprise, right? And you're running your entire company on Fable Fyber, whatever the frontier model is from an AI lab.

And then suddenly you know that the government can just switch the button off and suddenly your company can't do its thing. You're more incentivized to kind of run an open model at home that's privately influenced such that you can never shut it down. So if I was an enterprise that has been running Fable Fable Fiver has now been shut off. I'll be looking over at this GLM 5.2 thing and thinking, well, it's MIT open source. Yeah, maybe it costs 20K to run on hardware, but like, I'll rather spend that and save, you know, hundreds of millions down the line versus like going with Fable 5 and,

yeah, maybe achieving frontier level performance, but then, you know, being shut off potentially by the government, according to their agenda, like, that's not something that you potentially want. Now, I want to give a quick counterpoint to the whole Chinese open source AI models are going to take over the world because they're cheaper, they're as good, maybe not as good, but as good, good enough, right? Which is very simple. If you're an American lab that has a frontier air model that is expensive and you see your neighbors, or if you see your adversaries, China, distilling your model and presenting it as a cheaper model, you just do the same for your own model. And Anthropic has demonstrated that many times producing Sonnet. Sonnet 4 is basically a cheaper model of Opus

4.8, I believe. And then you see it with chat GBT with GBT Flash. These AI labs will produce a cheaper version and they'll distill it directly from their frontier models. And as these models get good enough to rebuild themselves, it gets easier to do. So I can see a world where they release Fable Six in the future with a companion model, which is like Sonnet 6, and it's super cheap for anyone that wants 85% of the capability and don't care about that extra 15%.

And it's super cheap. So it's competitive with the Chinese models. I don't think America has lost the kind of like cheat model argument, but the open source one, they definitely have. I don't see the American labs open sourcing anytime soon. Yeah, well, we saw Metapivot very clearly from the open source. Yeah. But like the savior of the open source world to close source very quickly. And I mean, that hasn't worked out too well for them or anyone really, which is disappointing.

There is a small caveat. Maybe we should cover about what open source actually means because it's not truly open source. There are still some secrets. I think a better way to classify. this is open weights. And when you go through training, there's, let's say, a trillion parameters. Each one of those parameters gets tuned over and over and over to reach training round, which happens trillions of times. And the output of this are the weights. It's just a large text file that has all of those parameters finally tuned that the model can run off of. What it doesn't

include is the actual source code that it took to make that. It doesn't include the ability to reproduce it. All it shares is the outputs. So while you could take their outputs and you could retune and fine tune those parameters to give you exactly what you want. It's not giving you the recipe. It's not giving you the secrets on how it built it. So there is still some proprietary knowledge as it relates to this open source model, these Chinese companies, because they are actually preserving the recipe in which they landed on this, the data that they trained on. There's a lot of secrets. The output is what's open source. And that's technically open weights. So when we say open source, I think what we really mean, whenever you hear open source model, chances are it's open weights.

And that's a pretty big distinction because that allows them to keep their kind of their secret sauce of how they do it. And it's also probably for the better because I assume you got to imagine they've been distilling some sort of stuff from, I mean, I remember seed dance. That was so like obviously stolen material because it was just able to reproduce all the copyrighted video formats from any public TV show in the world. So where they get their data from leaves a lot to be desired in question. But that's kind of the nuance between open source and open ways. And what we're getting right now currently is open weights. I don't necessarily believe it's open models versus centralized models. I think it lands somewhere in between.

Now, we've been noticing this new type of product that is getting used by a lot of software engineers and AI users. It's probably best demonstrated by this recent product release from Sakana AI. It's called this new model called Fugu. And they describe it as a multi-agent orchestration. system. Basically, how it works is you send their model a prompt as you do with ChatGBT GBT or Claude, and it disperses that prompt across many different models. It could be closed models like Claude and GPT. Could be open models like Jipu GLM or Kimi K2.7, as well as their own trained model called Fugu, I believe. And the result of this is like a gentic debate. So these

models kind of produce their own answers. Then you have another model that kind of judges these answers and produces the best answer from all of this. And the result from these tests is, basically, not only do you have a better quality output, but it's also cheaper. So the orchestration module basically picks the best models to do something when it's like cheaper and then only uses the best models

when it really needs to solve a really hard task that the other cheaper models can't do. So it saves you a bunch of money and we see it across other companies like OpenRaptor with their new Fusion API. The point being made here is we are headed towards a world where the ideal AI chatbot uses multiple models,

and they may not just be from the same company. So the question I have for like the United States government and any government that decides to regulate, whether it's open source models or close source models, how are you going to regulate every single model in the world, especially when the model labs come from other countries or are in fact open source. You can't regulate open source models.

That's the whole idea of it, whether it's open weight or open source. The whole idea is the government can't shut it down if you're running it on hardware at home. So it's just a really interesting nuance. I just don't think that the starts that the United States government has taken so far is necessarily the most productive one. I understand why they're doing it, but we need to figure out a different framework. It's funny because I saw this news this morning about this Sakana Fugu.

I think I'm pronouncing that right. I mean, surely, I've never heard of this. I don't know if you've ever heard of this. I think a lot of people watching him never heard of this company. They're Japanese. They came out of nowhere. And suddenly they're posting benchmarks that show that it has higher performance than Fable. And maybe that's true, maybe they use this mixture of agents.

But I think it's also notable that a lot of this is benchmarks. And I actually got some time to play around with the new GLM model this weekend. And while I'm sure it's great at coding and technical use, that's not really what I generally use the models for. And as I'm actually using these models, I'm giving it the general vibe test. I'm noticing that I really do strongly bias the American closed source models like GPT and like Anthropics Opus and Claude and, I mean, Fable, when it was available, it was incredible. And although the benchmarks show that it's very competent at coding, a lot of people aren't using it for coding.

They're using it for other things. And the general, the vibe check doesn't get past with these models, yet at least. So I think that's something worth noting, too, is like, these are just benchmarks. I encourage anyone who's listening, go try this out for yourself and see for yourself. Some people may actually get a lot of benefit from using a cheaper model. Some people just like having all the context in one place, and they want just a better overall experience. With the routing, I think this is a super interesting precedent that we're seeing with Sakana Fugu and how they are choosing to route their outputs through a series of

open source and closed source models in order to generate a better and more powerful outcome. I wonder the costs. I noticed that as I was looking through the documentation, there was no real costs associated. I have to assume it's not as high, but pretty close because it is routing through a lot of the private models and some open source models in order to get this, which means it's probably consuming a good bit of tokens. It's not totally going to be this open source, very low price model. But it is interesting to see this trend towards more router-based applications where not everyone needs to solve this incredibly difficult challenge.

Perhaps you spin off a few sub-agents. They use a more lightweight model to get you an answer without needing to consume a lot of those higher-cost tokens. So it's cool, innovative. I won't say it's novel. We've seen this before, but it's a new iteration of this that is now showing pretty compelling benchmarks. Yeah.

On the cost side of things, if it's anything like, open routers fusion API, which does the same architecture. It achieves roughly like 30 to 50% cheaper versus the frontier models, which isn't that major compared to like some of the Chinese open source models, but it still saves you a bunch of money if you're an enterprise using this at length. I'm trying to think about like the major takeaway that I have for myself after we've done this episode, Josh. And I think the main one is I'm inclined to say, and I hope I'm wrong,

that future AI model releases, fable and above, whether it comes from GPT 5.6 or 6 or other frontier AI labs, they're going to be more controlled in their release because governments are going to start getting more involved. We're going to start seeing nationalization attempts from different nation states in order to figure out how to release this AI models because if they're out in the wild,

they can exploit and cause some real damage. I don't want to think about what could happen in terms of a major event, but I think we're reaching that point where we need to pay careful attention. So that's what we're trying to do on this episode. At least that's what I'm trying to do. Yeah, I think that's right.

Like the speed and acceleration of these models and the cadence in which they're released is up only. If we had a chart that showed you the length of time in between major model releases, it is just getting shorter and shorter and shorter and that's not changing. So there needs to be a way to reliably be able to push these out. Otherwise, the gap between what exists behind closed doors and what's available to the public is just going to keep growing. And I'm not sure what implications that has,

but it sounds like it is noteworthy and something needs to change in a material way because the speed and velocity in which progress is being made is not slowing down. Like what does this look like a year from now? How quick are these models able to improve themselves? What are the benchmarks look like? Can we even create benchmarks anymore because it will be so capable? We're right on that cusp because we are approaching this

vertical asymptote off the curve. And it's just like it's a little weird. It feels like we're on this roller coaster and we're like kind of going down but I guess it's inverted where we're going up and we're going up really fast and you're not really sure. It's escaping. It's escaping control in a way. Well, I wouldn't say escaping control,

but it's just like that. It's definitely getting fast. And it's like, okay, like if you're driving your car really fast, you've got to be a little more careful once you reach high speed because like things, things can kind of get a little shaky quickly. So we're at that point. And models are getting very capable very quickly.

I can't imagine what Open AIs mythos class model looks. Like I'm sure they're working on them. We talk about, I mean, the hardware. I always think about, these are the Blackwell series models. What happens with the Veri Rubin series models? It's like, this, we are going to accelerate so fast. And I think it's important to, yeah, work on these safeguards now where it's still reasonable to catch up, where there's only one model release in which you have to focus on. And there's not 10 different ones from all these different companies

that are being pushed every single week. So interesting. That's the update. China is back with their open weights model, not to be confused with open source. And yeah, we still don't have fable access. So hopefully these things will get sorted, but I think it's, it's noteworthy that China, they never disappeared. I want to know what Deep Seek is doing next. I think that's my next question. It's like, where's deep seek at? Where's deep C V5 or V6? They just raised a massive round. $50 billion. That's their valuation. At least it's still a fraction of frontier labs. But yeah, they raised like, was it $9 billion? The founder himself put in $3 billion there. They're doing pretty well. And we haven't seen a model raising them many times soon.

Yeah. So we'll be fun to see. But that is the update on China, open source. Thank you guys so much for watching as always. If you enjoyed this episode, don't forget to share it with a friend who might also like the show, who might care about China or open source models, or wherever it may be. If you listen on a podcast player, rating us how you believe we deserve to be rated is always appreciated. We love the five stars. Those are always great newsletter twice a week. Next one is dropping on Wednesday, a day after you listen to this. And yeah, that's pretty much. I have one final request, Josh. It's something that you and I discussed on a on a warm.

last week. But we are in the market for sponsors or anyone that can support us. Please. We, Josh and I and producer Luke have been keeping the lights on this entire time and we've reached a point where we're feeling really confident about the numbers and all the support that you guys have given us. And we would love to have a partner that we feel very passionate about join us and support us in our vision of growing this into the leading frontier and AI tech podcast in the world. So if there's anyone out there listening to this that is inspired or wants to support us, let us know, DM us, you know, we're on X, we're everywhere. Just reach out and we would love to hear from you. That would be great. All the support is very much appreciated.

Keep the lights on around here and keep things going strong. So yeah, thank you as always for the support. If you made this long, you're a real one. And hopefully you enjoyed this episode. So thank you as always and we will see you on the next one.