NEW Chinese AI DESTROYS Mythos?

China's GLM 5.2 beats or matches Anthropic's Mythos so we're looking at a situation that just got released today that actually states that the band model Mythos and Fable 5 has already been matched by China's ZAI so they own GLM 5.2 and this was announced by the Wall Street Journal so according to the Wall Street Journal China has matched Anthropic in cyber security resetting the AI race. Now there's so much more to it which I'll come on to in a second but this is matching in certain benchmarks such as exploiting cyber security bugs which is pretty crazy stuff when you think about it because bear in mind like this is a fully gated model like nobody can access Fable 5 apart from like a small group of cyber security firms but we're looking at China at this point with an open model ZAI and GLM 5.2 matching Claude Mythos at discovering software vulnerabilities making a huge leap in the Chinese and US frontier AI race so apparently according to this as you can see this tweet by the bureau China's ZAI can match Mythos and it achieves Mythos level vulnerability detection. Now when I was testing out GLM 5.2 it certainly scored better than Opus 4.8 on GoldieBench really powerful model but also this is number one Chinese and number two it is an open source model which means you know if you've got a great setup number one the weights are open and shared and hugging face but number two you can host it locally if you have a great setup and so China have matched Mythos and this is something that I've always been saying especially recently is like the model is never the most an open Chinese model just matched Anthropic's most locked down system at finding security bugs and it's obviously for like a quarter of the cost of that so the Wall Street Journal called it a reset of the AI race and here's what you can learn from this so we've got Claude Mythos which is fully gated never actually released publicly this is the cyber security version and you've got Fable 5 that was released but then taken down by the US government and they basically said you know nobody can access to this right now and it's being released to like a tiny small group of people maybe it'll come back but there's no guarantees then you got GLM 5.2 from ZAI Chinese open way free to download matched a quarter of the cost if you're using the API an open model anyone can download just match the most locked down frontier model at roughly a quarter of the cost per token now this story was just announced a few hours ago so it's brand new this is from the Wall Street Journal as you can see so China's matched Anthropic and this is creating really an advantage for China but they're catching up I'm probably going to overtake because bear in mind like GLM 5.2 as well it's a self-learning model with self-iterating model which means GLM 5.3 will be a big step up again and it only gets better from here like Chinese AI is the worst it's ever going to be so if we look at the news you know Miphos is Anthropic's most powerful model it's good at finding security holes in software particularly in fact so good that they keep it locked away and only a vetted handful of US agencies and companies can touch it and this week China's ZAI said its model now does the same job finding the same security bugs matching the most guarded model in the world and its model GLM 5.2 is open anyone can download it and they can use it for about a quarter of the cost per token if you're using the coding plan with GLM 5.2 you can plug it into any of your agents and it's very affordable I've never run out of tokens on that model so pretty wild stuff also something that's interesting that Anthropic have come out with recently is that Chinese models are basically distilling and copying the models and there's basically rumors of China like backdooring the model and that was floating around even before Fable 5 shipped so whether you you know believe that or not it's pretty interesting stuff this whole race is really heating up right now there's also a different situation going on so China didn't just like you know they didn't copy a model they've actually built their own platform so a second Chinese firm called 360 security has shipped its own cyber AI the same week making the same claim that they're beating Miphos and they're building like a whole stack so this is like a homegrown rival to Miphos and they've actually got receipts of that so there's one model that auto discovers software vulnerabilities and a second system that runs cyber defense and that is China 360 as you can see right here so there's two things going on here there's GLM 5.2 and there's China 360 now bear in mind like the whole case for big US labs is that the models are moat right the idea that their models do things no one else can they're closed they're going to IPO probably going to be some of the biggest IPOs you've ever seen in history from Anthropic from OpenAI from SpaceX but at the same time if a free open source model matches them that moat gets very thin fast and also look if you look at the dynamics like both Chinese models are open way free to download and run whilst the US models stay locked down so when a free model matches frontier cyber performance the premium pricing behind US AI subscriptions and enterprise deals loses its justification companies already swap into cheaper open Chinese models just to cut costs so the model stops being the thing that you pay a premium for also who knows what's going to happen to US AI stocks next week that will be pretty wild as well I think we could expect some big announcements here but basically the way that I'm looking at this is like it doesn't matter what happens next so we have the agent operating system we have Claude inside there we have Hermes we have GLM 5.2 we have Kimi K217 code as well that just came out from China right and we can use them all inside our agent operating system and so the way that I would recommend you looking at this as well is that no matter what happens so if you know GPT 5.6 gets gated and then Fable 5 gets taken down but then GLM 5.2 is better and then you've got free models as well that are local you plug them all into your system you swap and change whichever model is going to win it can plug in there so models get gated they get copied they get pulled they get matched the system you build around them is the part that nobody can take from you and it runs on whichever model is the best right now because like both sides now hold the same superhuman tool right MIPHOS it wasn't a locked away for fun it can find the kind of security holes that take down like power grids and crazy stuff like that so that's why it sat in a vault tested on classified jobs kept away from anyone obviously that is not 100% confirmed but that's a risk that they're looking at and that's why it's been locked away and why the public can't access it but China have matched it anyway without even seeing the code and now like if you look at this what's scary here is like the most guarded cyber AI in America got matched from the outside which means the defensive tools protecting grids hospitals all that sort of thing well they've just been matched by AI which is Anthropic's most restricted system now this is just one point of view from this like you know judge from the outside there's nothing confirmed here I mean GLM 5.2 have come out and said it I wouldn't be surprised by this but I wouldn't like you know make any emotional decisions from it the main thing that I would say from this is like just focus on the system not the models now I'll give you another example so like inside Hermes agent we have mixture of agents which can actually outperform on many benchmarks Opus 4.8 by a long way and also Fable 5 right and so we have the system where we can get like frontier level intelligence even if it's gated in other ways because we can have our agents working together and so when you've got systems like that it doesn't really matter which models are pulled on and which models are pulled off so if you look at this like the US has pulled its own best models off the field Fable 5 got taken offline Mythos is restricted to a vetted view GPT 5.6 is still stuck in preview America's best models are sitting on the bench or behind a velvet rope meanwhile China's best GLM 5.2 is open free to download and already matching the frontier so the closed side keeps gating and the open side keeps shipping and that's pretty interesting this might be a gap that never closes Andrew Curran actually posted this tweet that was super interesting basically talking about how this won't be the first time models are pulled and we might be entering a new reality where we don't have access to like the latest frontier models from the US you know this is a new era really when you look at it it's a very messy one also some people are saying that the reason Mythos got pulled in the first place and this is unconfirmed but it was a report from The Verge that reported that a China-linked group may have had access to Mythos and that's possibly what triggered the Fable 5 takedown on Friday not this Friday by the way it was longer ago than that again wasn't fully confirmed but pretty interesting stuff so the plot gets thicker interesting things going on anyone can access GLM 5.2 right now for free we've already built it inside the AgentOS the main thing that I would say is just step back from the politics and everything else and just look at what actually happened so the single best model on earth the one government locked in a vault got matched in months by something open something cheap and it's not really a China story it's just a true story of like every model right models leak models get copied models get gated pulled beaten and made cheaper by someone else so the best models are lead measured in months not years so if your whole plan is like use the best model you're standing on sand and the ground will keep moving the thing that doesn't move the system you build around the model right so for example the memory that knows your business like we have an obsidian memory plugged into our agent operating system here it doesn't matter what happens right we're always going to have a great memory that's super powerful for our agents and our context the same for example with models whatever works right now we can plug this in we can switch the agents the workflows to turn a model into an outcome that's yours and that's what improves everything single day nobody can gate it or pull it offline for you and so you can learn this the easier way which is don't focus on the model any model can work with this focus on the memory the agents the systems the automations and focus on the outcome so for example if we've got like a video system like this i don't care which model we plug into this site we can create amazing models amazing outputs as you can see right here there's an ai avatar video of the b-roll the edits research the script we don't need gpt 5.6 or fable 5 or glm 5.2 even to do this we're just going to plug in whatever works right now get the job done so i would focus on the outcome and the system not the model itself you don't need mythos you just need a system and that's pretty freeing in a way because for example we were never going to get mythos anyway it's a lot to have vetted for you and waiting for fable 5 or gpt 5.6 probably a waste of time as well you don't need it right they're open cheap models glm 5.2 and its cousins already good enough to do real work the frontier you're allowed to use is plenty what's been missing isn't a better model it's a system that turns any model into an output for your business and that's why our agentic operating system is because we've got one dashboard that plugs in whatever model you want the gated frontier when you have it you open chinese model when it's cheaper a free local model for the everyday work and wraps it in memory agents and workflows that know your business whilst everyone else just fights over who gets which model you build the thing that's actually using all of them and you might also say like doesn't this like cost a fortune for tokens you can actually use free apis you can use the clis you've already subscribed to we've got claude cli plugged in there and for hermes like i just use my twitter subscription right and it can create amazing outputs with that so you don't even need to worry about tokens for this sort of thing but the main thing to focus on is like never feel behind again because everyone else is going to wait for the next best model to drop they're going to worry about all this china vs usa things they're going to be paying the premium for new apis and they're going to watch it get matched or pulled in months whereas with you after you've watched this you're going to own the system right you're going to build one system that plugs into any model use a gated frontier open chinese or free local whatever you want you pay near nothing for the everyday 90 and you swap a model in the day a better one ships and then you keep the memory agents and workflows forever so like for example the day that glm 5.2 came out we plugged it into our system and it was absolutely awesome we'd already tested it with our agent operating system so the point is that it's very flexible very adaptable and whatever happens you've got a great system that gets the work done for you so if you want the agent operating system from me you could build it yourself but if you want my setup where you can plug in any model you've got the local hermes engine agent kanban the cloud workspace every cli you already pay for in their free local models they are a mastermind group where agents can debate the memory involved token optimization playbooks and the systems where we use panels of agents like fusion sakana and even for example mixture of agents that achieve better outputs than opus 4.8 then you can get that inside the ai profit boardroom link in the comments description or go to the air profitable.com inside the community ask questions i answer questions with video tutorials every single day inside the classroom you get access to all my new best trainings so for example you can go from beginner to expert in just six weeks with this system inside the new daily update section you can get the latest daily update on the agent os what video tutorial on how to install it a step-by-step guide and also the resources on using this stuff and then we have new daily tutorials as you can see right here now inside the calendar you can jump on weekly coaching calls share your screen ask questions meet other members inside the map you can meet people in your local area and this is all available inside the air which has one single goal to help you grow learn and scale with ai automation you might say as well like isn't this stuff technical to set up actually if you look at our 191 pages of testimonials you'll see that many people have set this agent operating system up they're getting amazing results of it as you can see right here so for example shin was saying the agent os system is by far beyond the worth of the value it provides and also for example hockey said he set up the agent os said the system is amazing rick said that he just joined the school 30 minutes in and already built the os there's so many people getting amazing results with it so hope to see you inside there i'll see you on the next one cheers bye