that little company known as Anthropic, of course, Claude Code, their famous tool behind it, has caused quite the ruckus. Okay, we're we're talking about people are very very upset. Okay, I I I swear I could still hear Twitter people just screaming right now and it's been days. And if you have no idea what I'm talking about, honestly, you're living a blessed life. Okay, I want you just to look out on that farm and I just want you to smile because you you have it peaceful. you have a nice time, but the rest of us have to live with what happens out here in the real world. Okay, punk. You don't know what I have to see. But since you're watching this video, you will now discover everything. Okay, we're going to go over everything. Uh there's a lot of people yapping about what's going on with Anthropic. Uh I'm going to give you hopefully what is the full story, but more I'm going to also just I I'm going to make a couple, you know, I might project a little bit. I might make some guesses myself at the end. Okay, for those that are in the know, you've probably seen the image. For those that aren't, you have not seen this image. But effectively it says this credential is only authorized for the use with clawed code and cannot be used for other API requests. This is the big image that kind of kicked off this whole anthropic hate that's going on right now. So what exactly is happening here? Well, if you don't know, Anthropic does have plans. You can get a pro plan, a pro 5x plan, and a pro 20x plan. And if you're on that plan, you at least up until now could use that token and say cursor. you could use it inside of Open Code and be able to use your plan and use your set amount or limited tokens to be able to do a bunch of code changes. Well, that all changed this week. They said, "Hey, guess what? Absolutely not. Nobody else is doing that. You can only use the plan and claude code." Now, this came as a big surprise for everybody. Okay? People were not happy. The GitHubs are just, as you can see, just filled with just people trying to find workarounds with it. Tons of comments, ridiculous amounts of comments going on in there. people saying they're cancelling just a lot of stuff lot of people not happy so in other words if you pay for a subscription plan you cannot use thirdparty tools you must use cloud code if you pay by the API pricing which is literally like something like 10x more expensive then you can use things that are non cloud code you can use open code you can use cursor you can use whatever but there's a lot more to it I think this Reddit post right here is honestly really illuminating uh they're effectively saying that everybody's is kind of lying about the situation. Okay, people aren't giving the full truth or at least they're giving maybe a partial truth is probably better. I mean, they are saying, "Hey, yeah, there's misinformation going on." Well, it's it's more like partial information. It turns out part of Claude Code's uh big plans is they they actually explicitly state in their toos, this is only use in our products. And so, has this always been known? Yes. Apparently, this have always been known. Effectively, what a bunch of these tools do is that they use your OOTH token and they forge requests with specific headers and body. It's very fragile. Like if if you've ever done this before, uh have you ever had to treat like a blackbox service and you have to kind of guess how it needs to behave and then you get it up and running and then they make a change and it just just destroys your service until you come back up with the new change that they've made. It's always very fragile, super frustrating. But this is how all the tools work. And this toos thing, it's been around since all the way back when Claude Code released in early 2025. They have always strongly discouraged using their APIs this way. They put real effort into preventing it. So, in other words, this problem that everybody's very angry about was actually just really a ticking time bomb underneath because Anthropic didn't enforce it. They allowed it to continue to fester for a long time and a bunch of people switching over. Don't worry, we're about to like literally we're about to get the tin foil out. We're just we're not there yet. I I want to make sure you guys really understand what's happening before I go off and yap project. Lastly, apparently the person running Claude Code has came out and said this on Twitter, which said this. Yesterday, we tightened our safeguards against spoofing the Claude Code harness after accounts were banned for triggering abuse filters from thirdparty harnesses using cloud subscriptions. Third party harnesses using claw subscriptions create problems for users and are prohibited by our terms of service. They generate unusual traffic patterns without any of the usual telemetry that cla code harnesses provide, making it really hard for us to help debug when they have questions about rate limit, usage, or account bands. And they don't have any other avenue for the support. In other words, it's just that these these abusive third-party platforms are just impossible to debug and any problem is perceived to be an anthropic problem, not the tools problem. And so that's what they're really saying. They're they're really just saying, "Hey, for your best interest and for our ability to debug, the only possible way forward is for us to make sure that we're always uh you know, enforcing our toos no matter what." All right, so that's kind of like the what has happened. All right, everybody, we are now on the same page. Now, the real question is why did this happen and why now? Like the Open Code has been using Anthropic for a long time. cursor has a lot of these tools apparently like cloud something or another that's some other service is also using it like this has been around for months upon months upon months so why enforce it now especially at the height of clawed code glazing the first thing people kind of talk about which honestly I think is is is the weakest and probably the most incorrect of them all is the cost right so remember there's three potential plans there is going to be a pro plan and then you have the 5x plan and then you have the 20x plan. This one's approximately $20 a month. This one's 100 I believe and this one is 200. And so often what you see people kind of citing is that okay people don't really make usage of the uh pro plan except for maybe just a little bit actually cost more and then the of the 5x maybe half the people cost more than the actual thing and then for the 20x the large majority of them end up costing a lot more and to what extent they cost I don't know. So the general calculus everyone runs on the math to like oh this is why they're doing it goes something along the lines of this right the integral of p over whatever its bounds is I'm not going to write its bounds uh plus of course the integral of the 5x plan plus the integral of the what's it called of the 20x plan is going to be like minus the integral let's just call whatever this line is I'm just going to call it uh g of x because I have no idea what the hell it is let's just call it g of x this thing has to be greater than zero and that's why they're doing it right the area underneath this curve is bigger than the area underneath these little price point stepladder curves, right? Like that's what everybody's saying. Everyone's saying, "Okay, this is the real cost." That's that. Honestly, I don't think this is the problem. I mean, yes, I think money is the problem. But I don't think this particular thing is the motivating factor or why we're seeing it now. So to kind of put us in the right direction, I believe that DAX actually has a really good like at least directionally correct tweet, which is okay, it's their business to have the right to enforce their terms however they like, which is true. Like they already said from the beginning they don't want anyone bug cloud code using their subscriptions, so it technically is their right to be able to do this. They're not uh obligated to give anybody access to their services in any other way than they have set forth in their terms of service. And if you've paid for it and you are now upset about it, you don't have to continue to pay. just stop paying. So Dax's projection is that models aren't sticky enough. Yes, Opus is good, but so is Jippy 52. There's things that Jippy 52 can do better, and there's things that Opus does better. So, it's not like the models are actually that sticky. What they really want is that they want their entire tooling, their entire stack to be what you use. When you use Claude, you use Claude code. This way, they own everything. They show that they have more value effectively. And I think this is a big thing, especially when it comes to Open Code, because you may not realize this, but Open Code has been growing wildly. They're about to hit 1 million monthly active users on their open-source project, which means before it was just a toy project, but now that it's just shooting up so wild, and the reason, by the way, it's shooting up so wild. If you have not used cloud code, like I'm not trying to be mean, Claude, but I feel like I kind of have to be mean because let's can we all just agree, Anthropic is the number one fear-mongering, hey, software engineering is going to end company of them all. And Claude Code the product is just significantly far behind open code. They've been having screen flickering just like it is like a dance club in your 2e editor that is honestly just like I'm going to get epilepsy from even looking at it. They've had this for months now. They tweeted about how they fixed 85% of it, which is an absurd fix. That's not a fix. You don't know what the hell's going on if you can only fix 85% of it. And second, how did you even come up with that number? And third, one day later, they had to roll back that change because other bugs were cropping up. What it shows to me is that the company in which is telling you that software engineering is over can't even run their own like their own software. They're not making the changes that are needed to be competitive and they're losing out to a bald man and a vegan from the Ozarks. It's just diabolical these two. Yeah. By the way, you never you everybody knows you don't start a land war in Asia and you never go uh headtohead when open source is on the line with the vegan. Okay, everybody knows that. And by the way, even with cloud code, besides for the flickering, uh when you type in Ultraink, it's like this rainbow. I had I had it more than once shift off and have like the antsy escape sequences just being printed in plain old regular text. I mean, just just egregious bug after egregious bug. I couldn't expand co code with control O while in a menu to accept the code changes. So it's like, okay, accept these code changes, yes or no. Well, if you want to see the code, expand it. Well, you can't leave the menu. You're just like, oh my gosh, please. What are you doing this, Claude? Stop it. Stop it, Claude. Just let me Oh, I'm going to The tool is not good. Okay, I'm first off, an impressive software project and they've done good so far, but it is not that usable comparatively to Open Code. Open Code is a beauty. Okay, it's absolute beauty. Gives you all the information you know right up front. You It's actually welldesigned. It's by developers who like to code for developers who like to code. And it's obvious. So to me, when I see this graph and people talking about this, I don't think so. I don't think that's it at all. All right, people. We're going to need to do something. I'm going to need the tinfoil hat up here for a second. Okay. All right. So this is what I think the real problem is that Anthropic is actually having. Okay, I don't think it's just simply that previous graph. What I think it is, it is partially that previous graph right here. But I think there's a lot of lies in that previous graph. Okay, I think the amount that they're subsidizing uh the Pro, the 5X, and the 20X is much more than this graph leads you to believe. I think the cost that they actually have to run everything is significantly higher. First off, you got to remember they have to do these huge amount of training runs that involve buying and being able to utilize a whole bunch of hardware and energy which is just going to be a gigantic number. Then you have the whole lifetime problem of the model. How I mean who here's out there still using 35 when you got 374 and 45 on the table right now. You also have Jippidity 5, Jippidity 51, Jippidity 52. You have all these other options that are better than 35. at 35 costed a huge amount of money. And so the lifetime like operation of these models, especially right now in such great change, is rather low, but the cost to actually run it is rather high. Plus, you have employees which are extremely expensive. It's stock grants that some of these higher paid jobs are just getting paid outrageous amounts of money. Plus, you also have to have all the infrastructure and all that crap being set up to be able to run all the stuff that involves all the continual running of energy replacing GPUs and all that. And then on top of it, you got things like Nvidia just dropping brand new AI GPUs called Reuben that apparently is going to deliver 10x in reduction in inference token costs and 4x reduction in number of GPUs to train models. That means MO models. Okay, when you want more models, you got to get the new chip. And so this is like this constant rat race of trying to stay ahead of all the competitors like Chat, Jippidity, Google, XAI. I mean, let's just I mean, I know XAI really is only good at putting people in bikinis when they don't want to be put in bikinis. But one day, you never know, okay? You just never know. So, there's a real subsidation like subsidization problem going on here. These tokens aren't just subs, you know, subsidized by like just the pro users versus the 20x users. They're not subsidized by like 80%. I would not be surprised to find out that these things are like 98% being subsidized right now. Meaning the true cost of the max plan currently is 50x more than they're actually saying because they're still trying to recuperate all the other things beforehand. And then to remain competitive in the future, there is a continual non-stop march into every single latest piece of hardware and everything that they must put in there. RAM RAM has already been purchased for all of 2026. So what they have is what they get. And it's astronomically expensive with Samsung being like, "Yeah, we're going to just we're going to just 4x our price. You going to pay for it?" And they're like, "Yes, absolutely. Please, yes, sir. Uh how when would you like me to bend over and how often?" Right? like they're just absolutely at the whims of everybody. And so if they can make a stack that's so competitive and then force people into this highly subsidized plan that can only be used on their stack, their stack becomes the sticking point, right? Because you can't use the cheap effective access to unlimited tokens. And then on top of that, you don't have access to other models when you don't want to use them, right? That's what that's why open code's so dangerous. It isn't because you can use your Mac subscription. It's because when I go, "Huh, let me try Gypy 2." Oh, wow. Jippy 52 did really good right there. Maybe I don't want Opus. All of a sudden, yikes. Anyway, so there you go. That's my thoughts on the whole thing. I know maybe I'm a little bit more of a fringe conspiracy theorist when it comes to this stuff. But I think that I I I personally see trouble in the water. Okay. I think that that's why they're doing it because why would you torpedo the single highest amount of developer gratitude and non-stop Twitter glazing? It's so over. We're never coming back software engineering. Opus is right. I mean, what you've seen videos on saying Daario was correct that 90% of code is going to be written in 3 to 6 months. I mean, it's just feverpitched. We're talking about absolute feverpitched zealatry going out there on Twitter. And then they just take they literally just take old Yeller out and they bam. It is just insane. So there you go. That's it. That's all I have to say. Hey, I'm going to try something a little unusual. You want to press like. You want to make a comment? Tell me where you think I'm wrong. Tell me where you think I'm right. I never usually invite these kind of things cuz honestly I think it's kind of I think it's kind of weird to do that. Yeah, I know. I did push for subscribers for a little bit to get over a million. By the way, thank you for the million. I appreciate that. Anyways, thanks for watching. The name is I don't trust Daario. I don't know if you know this, but you may not know this, but Daario, guess what he said? He thinks that open- source models are a danger. He's always pushing for regulation. He does not like open source. He DMCAD a bunch of people when it came to them accidentally leaking their code and people deoffuscating it. He does not like open source and Anthropic doesn't like open source and they want to own everything and they think that AI is too dangerous for anybody but them to run it. And they ultimately just want control over you and your life. They want to leave you helpless and incapable. That is why 90% of all software is going to be written. 100% of all software is going to be written because they don't want anyone to actually be learning and independent. They want everybody to suckle from the teeth of Daario. And that is that and that is what I actually believe. I think that Daario is one of the most dangerous man on all of the AI landscape and that I don't think he has your best interest at heart. A gen. I know it's not easy to find developers to work with. Trust me, I hired TJ. So, how do you find quality people to work with without spending all of your time on LinkedIn? Well, with G2I, you can go from interview to first PR in as little as 7 days. G2I has helped people get placed at Meta and One Password, and they run React Miami, so they're super embedded in the JavaScript community. So, if you want to find topt talented software developers, check out g2i.co. Cow.