Gemini Argon 4 is HERE!

Julian Goldie (0:00)

So today, we have a brand new update from Google Gemini 4 Gemini 4 Argon has dropped, and I've already plugged it into our Gentic operating system. I'm going to show you what we've built, how it performs on benchmarks, what we've created so far, and it is a very solid model from what I've seen so far. It's created some pretty awesome stuff. There's some things where I'm like, it's okay, but you can see, for example, you can create a full 3D worlds. It can, for example, create 3D games. It can build all sorts of cool stuff. We're going to show you exactly how that works in a second. The other great thing about this is, if you're already on the Gemini subscription, the Google AI Pro plan, then you can just plug that CLI directly in to an agentic OS like this. So we're using anti-gravity as a harness. And then we've got Google Gemini for Argon plugged directly inside here. So if we just check it's working.

So before we come on to the benchmarks, let's just have a look at what we've built here. So, for example, we created like this full 3D world, as you can see, and I'll show you how that compares to Opus 5.5 in a second as well, which is obviously Anthropix competing frontier model side by side. But it can build some interesting stuff. Let's have a look at the next one. So this is Paper Kingdom, which is like a 3D sort of design of a flip book that you can use as well.

If we go back to the gallery, what else we got here? We got Pocket Planet. So Pocket Planet is a 3D world. Now this one is kind of strange to me. Now again, I'll show you in a second how this compares versus O plus 5.5. But for some reason, he's like swimming around the planet instead of walking. And then he starts standing up and walking, but he seems to be moonwalking there and walking backwards because the controls are completely backwards. So it's good and it's not good. Like you might be thinking, okay, is it better? Is it worth switching? Should I be using Gemini? And I'll come on to that in a second and the best ways I think you should use it. But it can create some pretty cool graphics as you can see. Let's view the next one. This is Voxel Island. It has this nice style when it's designing. I don't know about the details. That's one thing that gets me on a lot of these models that claim to be frontier. And then when you test them out, you're like, yeah, I'm not sure about that. And so one of the things, for example, when I was testing out Muse Wonder yesterday is like the detail was not that nice. Now this is one of the most interesting things that it created, as you can see, so it can create like these beautiful animations of different shapes that were plugged in here, which is just nice to see. Okay, how does it think in terms of logic? How does it do in terms of design? Obviously, you're not going to be creating that for your business day today, but it gives you a good feel for how good is it, UI, design and everything else. And then it created this website, which I actually thought was pretty mediocre, pretty mediocre. When I look at this, I'm like, we could build better stuff than that for sure. And again, I'll show you some examples of what we built with Opus 5.5 and everything else. So this was like a 3D website that it built. There is something with the design that I just don't like. This one is quite nice, but at the same time, it still feels kind of basic when we're using it compared to something like Opus 5.5. Now, if you're wondering, okay, what are the best ways to use it? For me personally, I have it inside the Anti-Gravity CLI, and I've plugged that into our Agentic OS.

And then you can see it's working inside the Anti-Gravity CLI here with Gemini 4 Argon. And we can have that working alongside all our different models for Gemini, as you can see here. And we've got all our builds inside the preview section. Plus, anything that we build here, we could give to a different agent, either deploy for us or to improve, or that sort of thing as well. So I think that's one of the best ways to use it. And also we've got it inside a beautiful Mission Control, where we can have all of our agents in one place, and it's super easy to manage. Now, you could also use this with other agents, I'll come back to that in a second.