Meta Muse Wonder DESTROYS Opus 5.5?

Julian Goldie (0:00)

Muse Wonder just dropped today. So there's two new frontier models from Meta that have just dropped today. That's Muse Wonder and Muse Wonderful. These are two different models. You can see the differences right here. So Muse Wonderful and Muse Wonder. Now this is only available for Muse Wonderful on Muse Code subscription. If you want access to Muse Wonder, it's available on the API, and you can also get it on OpenRouter via the API. Now, if we have a look at the benchmarks, let's compare them and see how they're performing. If you actually look at the benchmarks here, it's compared against Claude Opus 5.5 Max, and you can see here that Wonderful is more powerful than Muse Wonder. Now, if you're wondering, okay, how does that perform in reality? How does the creations of Muse Wonder look like side by side? I'm going to show you some of the stuff that were built side by side, and let's pull these up so you can see the differences. So we've got Muse Wonderful and then we have Muse Wonder, as you can see right here. Muse Wonderful is a little bit more interesting, a little bit more, it creates some interesting designs. When I've been creating, like encoding with it, and it only just dropped today, so we've just built a few things with it, but so far from what I've tested, it does look a bit more interesting. Whereas, for example, when I actually look at the creations from Muse Wonder, it's a bit more like a kind of style where, you know, it feels a little bit like Grok 4.6 to me or 4.7 when I build with it. I don't know why that comes to mind, but how it works step by step. If you're wondering, okay, what does it look like on the full run, the first run? So this is the one shot example that even with Wonderful, it didn't work first time. Whereas, for example, if we have a look at Claude Opus 5.5 here, we can see that side by side, the designs from Claude Opus 5.5, bear in mind, this is just one shot. So it was created in one single go, one single prompt, and Opus 5.5, I think, is still superior, despite what's on the benchmarks here. And I think that's really important to note. Now, we've already plugged it into Muse Code, and what I've actually done is use it inside Muse Code, inside our agentic operating system. So I think that's one of the best ways to use this because for most people, they're not going to use as their number one coder or their first agentic harness. But if you've got a system like this where you can have it side by side and then in one click, you can go to Hermes or you can go to Claude, I think that's the most fun and powerful way to use this because even though it may look great on the benchmarks, Muse from Meta is just not as good in reality. If we have a look at this to do list app, as you can see right here, it's cool. You can build some cool stuff. Don't get me wrong.

And it feels fun to build with. But at the same time, it still feels a little bit retro, a little bit just not that smooth. And this was built with Muse code, with Muse Wonder plugged in. And we can switch between them. So we've got Muse Wonder 4 over here, and then we have Muse Wonder, and those are both plugged in to our Gentic Operating System. And then the other cool thing about this is we can have a look at our previous conversations with it. And then additionally, if we build anything, we can save the build and it goes straight into our workspace over here. So I think that's the most powerful and fun way to use this. And bear in mind as well, this all plugs into our memory system so that we can get access to Muse Wonder and our memory system from day one. So if you've not been using Muse, which 99% of people haven't been, then having a memory system like this, a memory galaxy that syncs across all your agencies is one of the best places to work with that. And that's inside our AI Profile Board if you want to get access to that directly. So that was the first thing that we built with it. If we have a look here as well, we can have a look at Skyward Isles, which is like a floating islands simulation. So let's have a look at this. We can look at Muse Wonderful after it's been fixed versus Muse Wonder after that's been fixed. So both of them took a bit of back and forth with this. And you can see the examples right here. So Muse Wonderful, it does feel a bit more detailed, a bit more interesting when you look at it side by side like this. I think it's a lot better at coding, a lot better at frontier design.