They Just Shut Down Fable 5 + Mythos (Do THIS instead)

Guys, it's official. We can no longer use Claude Fable. I hope that you got some good building in. I personally, for 3 or 4 days straight, was building with Fable, and I have to say it was very, very impressive. But, in today's video, I want to talk a little bit about what happened. So, yeah, this has been some pretty massive news. This is the first time, as far as I know, in history of AI that yeah, basically a model has been taken away from us. This is not happened before. Uh the US government basically has told Anthropic that I guess Fable is just too strong. I have to say, personally, I don't actually massively care about this. First of all, I think they'll come out with something very, very soon. Second of all, I wasn't really waiting for this model anyway. I'm actually waiting for Haiku 5 or something, like an improved Haiku model is actually what I'm waiting for, which might sound weird, but So, like, this is kind of my sentiment as well. After all that, the results weren't noticeably better than Opus 4.8. Maybe I just didn't hit the right use case. There are some interesting comments. It identified a simple but deep ordering issue on my ML training pipeline. Fixed some cases where I was unknowingly using format methods in some cases, right? And I have to say, Fable, like, unnerfed Fable was running for an hour, an hour and a half sometimes doing certain things. I actually I set it a goal the other day. The goal was basically to make uh this here. I can't show you cuz obviously there's people's names and stuff, but basically, if I open this, it's a CRM, right? So, what it does is it shows the leads on this project that are hot, right? This is something that we've been working on for a while. But, like, there's a lot of different things happening here, and basically, it was able to run for an hour and a half on slash goal. But, yeah, just going back to kind of what the alternatives are, I'm probably just going to be using Opus 4.8 going forwards. There have been some model drops today. So, there was GLM 5.2, but they're not, as far as I know, they they haven't released this on open router, which I'm not really sure why the hell not. Um I don't know what they're playing at. Yeah, so they haven't released on open router. Just get on with it, honestly. Nobody cares about some blog posts and stuff. Like, come on, get the model out to people. I've seen some tests of GLM 5.2 already. It seems, you know, pretty standard. Uh I have to say not many max, what's it called? Um Gimme 2.7, the one that I tested yesterday, is extremely good. If I were on a budget right now, I'd be using this as an alternative to Claude code. Gimme 2.7 code, also just Gimme code generally, is extremely good. Like, the fact that it says that it's even close to 4.8, like, I might actually not even disagree with this necessarily, because the thing that I produced yesterday when I did this test was extremely good, right? But, if you want to know what I'm using going forward, I will be using Opus 4.8 inside Claude code. Nothing has changed. I've talked about this many times on the channel. In my opinion, we already have all of the tools needed to build something. If you're waiting for Fable 5 to build something, then you probably just need to learn like or like fill in some gaps in your knowledge, is what I would say. Like, I was able to build this incredible project, harborseo.ai, um with, you know, Opus 4.5, for example. So, yeah. Basically, what I'm trying to tell you guys is you don't have to worry too much about Fable being gone. Yes, it was a good model. I'm sure we'll get it back as well. In my opinion, this could also be kind of um the US government taking revenge on Anthropic. It could be Anthropic hyping um Mythos {slash} um Fable even more. But, like, you don't need them, guys. This Everything you're seeing here, which is a very, very successful SaaS project with quite low churn and a very, very dedicated audience, um you know, I'm able to ship new things pretty much every every week without Fable 5. The actual important thing is how you build your project, like the foundations of your project, like having a good database, having a scalable kind of um back end. Having some kind of safety, security, having a fast front end, having good SEO in the front end, um connecting all to Claude code, and then having an API that Claude code can control. You make all of these first, right? Small pieces, put all the pieces in place, make sure that it's a good foundation, make sure that it's well organized, make sure that AI knows what's going on. Put that inside a Claude code conversation, and then like once you've got that, like you can see here I'm doing um bug fixing here. So, I actually have a system where I can just get the feedback directly from the website, right? So, people give feedback, I just import it here. I'm using Opus 4.8. I haven't got any issues with that whatsoever. I'm not crying over Fable 5. It just doesn't matter to me. Like I was building happily with Opus 4.8. Sure, Fable was an upgrade. Sure, it ran for a bit longer, but I'm also sure that they would have nerfed it within 2 weeks, and we wouldn't have been able to use the real original model anyway, right? So you guys, I just wanted to give my two cents on that um because I think a lot of people are saying like uh you know, it's over type stuff. Fable That's probably going to be in my title, which is hilarious, but um you know, you don't need Fable, right? Another option, by the way, if you do want more options, is Codex. Codex is is Sure, you can talk about GPT all day and say that it's nowhere near Opus level, blah blah blah. I mean, I code with it all the time. I really like using it on the low setting um because it's just so damn fast, and I love fast intelligent models. I've talked about this a million times on my channel. Um if I can choose between fast and intelligence, you know, I would kind of want both. Whereas Opus is intelligent, obviously, but it's not very fast, right? Whereas low is uh GPT-5.5 Codex low is um fast and intelligent enough to do a lot of the things that I'm doing with it, right? So, again, Codex is 20 bucks a month. You get a lot of usage when compared to Claude. So, that's just something to bear in mind. And then, just before the end of the video, guys, quick shout-out for my sponsor, me. This is Harbor SEO.ai. It's a project I've been working on for an extremely long time now. It's a project I'm very, very proud of. I've just shipped Dev Harbor, which allows you to connect it to Claude Code, create an API key, and, you know, use your tokens within Claude Code. So, if you are writing or creating If you're creating a website, you can use Harbor to power your content. I have a lot of amazing features in here, including Co-pilot, where you can basically have a conversation with your Harbor data, and also get it to trigger anything inside Harbor. We have keyword tools, and we have writers. I just shipped an update yesterday making the cluster system a little bit better. It's now better at identifying potential clusters that will actually bring you, you know, traffic and clicks. And speaking of traffic and clicks, guys, the proof is in the pudding, as they say. We've done 1,888 pages published, 600,000 impressions, almost 10,000 clicks. And yeah, this is the founder pricing. Founder pricing is now back. So, you can get it for the next 17 days. Sign up today, guys. You basically get 50% more articles, if my math is correct. I think it's 50%. You also get six times rollover. So, if you don't use all your articles, they roll over six times, which is really, really nice. And yeah, I mean, you will literally not find a better um AI SEO content generator and also better priced AI SEO content generator. So, go and check out Harbor, guys. I'll leave a link in the description and in the pin comment. This is kind of your last chance to get the founder pricing. Uh, I probably won't be bringing it back after this. So, make sure you check it out, guys. And yeah, it's the best way to support me and to support my channel. And if you do have any feedback anytime, just press messages right here and write some feedback and then watch what happens when I press send. It appears in my admin dashboard and I will get to it pretty much within 24 hours every single time. Thanks so much for watching. If you are watching all the way to the end of the video, then you're a true legend. Go and check out harborseo.ai and I'll see you very, very soon with some more content. Peace out.