Qwen 3.8 VS VS Fable 5: Who Wins?

3.8 versus Fable 5. Who wins? I actually built over 50 different tasks with Quen 3.8, and I'll show you what we discovered and how they compare in a second to go. And also, if we compare them on the benchmarks here, you can see the Fable 5 is on a similar level on some of the benchmarks versus Quen 3.8.

But the difference is the Quen 3.8 Max is actually free, number one. And number two, it's actually going open source as well. So this is, number one, like a huge threat directly to Fable 5. And I think they've probably got to release something soon, especially with Opus 5 catching up.

But at the same time as well, I want to show you how it actually performs in reality because benchmarks and actual tests are very, very different. So if we have a look, for example, on Terminal Bench 2.1 for Terminal Agent, you can see here that Quen 3.8 Max is scoring 86.6 versus Fable's 86.6, right? So there's actually quite a big difference. And you can see here that in some ways, Quen 3.8 Max is actually outperforming all of the frontier models on the market.

How does it perform in reality? Well, that's the most important thing, right? So if we have a look, for example, all right, so if you compare this side by side, you can see our live builds. So these are Quen 3.8 versus GPC 5.6 Sol versus Fable 5 versus KimiK3.

Now, first thing you're going to see is this generation from Quen 3.8, super nice. Like SYN 3D, it feels nice to use, easy to use, and it's actually a step above all the other builds. Now, when we've tested this, this doesn't mean that like Fable 5 straightaways is not as good as Quen 3.8, but you can see the full power of Quen 3.8 and how good it actually is when it comes to building stuff. Here's another example.

So this is a kind of open world RPG game, as you can see, and the gameplay is smooth, it's pretty easy to use, pretty fun to play, pretty difficult and challenging as well to play as well. And if we compare that versus, for example, Fable 5, Fable 5 is like good, but it's a little bit slower, not quite as interesting straight off the bat, and also bugs out a little bit when you're moving forward here. Now, KimiK3 can also build super nice stuff. So if you actually look on front end design, I would say that the front end of KimiK3 in this example looks way nicer than the front end of Quen 3.8 Max.

Some other stuff like this Neon Racer game, again, like huge step up versus Fable 5 on this particular design, which is pretty nice. And I just noticed that like everything feels super fluid in 3D when you're building out with Quen 3.8 as well. Here's another example of like a 3D racer game. This actually, I wouldn't say is that good just because it's a bit difficult to play.

Once you get past that bit, pretty fun. By the way, if you want to check out these builds, you can check them out at goldiebench.com. We have all of our builds and all of our tests side by side with a leaderboard. So you can see which model you prefer and also play the tests and see what you think yourself, or you can test them out yourself with similar prompts.

The thing that I would say here as well is like, it's really good to test out games simply because it gives you an idea of coding, logic, how it plans, front end. It gives you basically everything in one single test, which is super useful. Now, if we have a look at this RPG game as well, pretty good. I mean, the camera angle is not that great, but again, really, really nicer.

Let's check out the flight simulator. Now, usually I've noticed for some reason, like KimiK3 and open source models just in general struggle with this. Now, if we have a look at Fable 5, Fable 5's version is probably the nicest, which is this one down here. You can see with Quen 3.8 that it's really struggled on the camera angle.

It doesn't really make sense. The controls are weird and you can't really use it properly. You can't play it really well. So that's the first time I've seen Fable 5 actually outperform Quen 3.8 on all our tests.

So have a look at this one. Now, this is really cool. Like the intro, as you can see, the pop-ups, it's a super fun game to play. Pretty crazy actually once you start playing it and the detail as well on it is really, really nice.

If we have a look at Fable 5, what you can see over here is it's not so good. The other thing that I would say there's a big disadvantage with Fable 5 is like sometimes it would just automatically switch over to Opus, which means like you can be halfway through a build and then you're like, well, why am I getting bad results? And then you look and you're like, oh, it's because it automatically switched over to Opus instead of continuing to use Fable 5, which blows my mind really because, you know, you could be coding something out for hours. It could be running on goal mode and then it just automatically switches over to Fable 5.

And this is also a lot slower with less detail to play, whereas a version with Quen 3.8, much more interesting. So we'll look at this Dragon Realm game here. So this one actually feels super basic. Like if you look at the way the character moves around, the graphics, et cetera, it just feels really, really basic.

If we have a look at Fable 5s, it looks a lot more interesting, but it's kind of like a, kind of like an old PlayStation game. I would actually say that of all of them, KimiK3 probably nailed it and GPT 5.6 Sol did a really good job here as well. Now for me, if you were to ask me like, which AI do I use day to day, I'm still using Fable 5. And the reason for that is like, it's a lot more reliable and consistent.

Whereas I found with Quen 3.8, you have to go back and forth with it a lot more. And also it's just way easier to use if you're building something big, like for example, we have our AgentOS and we have all of our websites plugged into Cloud Code. Number one, we've already got everything plugged in right there, so it's kind of hard to switch. But number two, it's far more reliable and it's easier to use.

So if I'm building something big, you know, that could be an app or a website or that sort of thing. I'm going to stick to Fable 5 for now. And I think they'll probably release a new version of Fable pretty soon as well. Whereas Quen 3.8, what I think it's really good for, in fact, perfect for, is like, if you want to use a frontier model, but you can't get access to Cloud Fable 5 simply because maybe you don't want to pay for the subscription or, you know, for some people, Cloud isn't actually available for them.

In some countries you can't use Cloud, then I would use Quen as well. And also the biggest difference here is like, with Quen 3.8, you can plug it into, for example, Hermes Agent, and you can use the coding plan directly with Hermes Agent, and it's designed for agency tasks. Whereas with Cloud, you can't plug that into Hermes Agent directly. They've stopped people from using the Cloud CLI inside Hermes Agent.

So that's another benefit to this too as well. And also bear in mind that Quen can actually generate images and videos, but Cloud can't do that as well, right? So if you want to generate images and videos, I'd probably prefer to use Quen simply because otherwise with Cloud I'd have to plug in a separate API, and then it's going to be super messy as well because it doesn't know how to prompt the AI properly without proper training. Also, if we have a look at this game, Fable 5 again, created a much nicer generation with more detail, more interesting gameplay, and everything else.

Whereas you look at Quen, it's quite nice, but I just, I don't like the camera angle so much, and there's not as much going on there. Now this one I actually saw a big difference on. So if we look at Fable 5, a bit buggy when we're using it, when we're moving around, et cetera. Whereas, for example, when we play with Quen 3.8 Max, like this game is super smooth.

It's less linear, easier to play. You can see how buggy, by the way, GPT 5.6 Sol is when we're using this. Side by side, I would still say Quen 3.8 consistently outperforms GPT 5.6 Sol when I'm testing out. But also you got to bear in mind as well, like when there's no real place to start coding with it properly, unless you use maybe like inside cursor or something like that, right?

But with Claude, you get the harness and you get the model itself, which is super useful. However, what you can actually do is you can use an agentic operating system like we have here. So we have Claude code, but with different brains plugged in. So we can switch to GLM 5.2 in one click, we can switch to Kimi K3, we can switch to Quen 3.8.

And that means we can use the power of Quen 3.8 inside Claude code, and then we get the best harness with one of the best and cheapest brains, right? The other benefit as well is like, you could use something like Kimi K3, which is almost on par as well, but a lot cheaper if you're using the coding plan, you plug that in as well, which is really, really good. But yeah, definitely seems better than Fable 5 in some areas. So have a look at this one.

This one, I would say again, like Fable 5 outperformed Quen 3.8. But yeah, overall, really cool model, super useful, can build some awesome stuff, and you can use it for free, and also it's open source. So like I said, I'm going to stick to Fable 5 for building big things, but if you want a frontier model for free, then I would actually use Quen 3.8. And if you want to combine both and get the best of both worlds, you can check out our agent operating system with Quen 3.8 plugged into Claude code.

And also the great thing about this is as soon as a new model comes out, and there'll be another model next week, let's be honest, then you can plug in your memory system. You've got all your systems ready to go. It doesn't really matter about chasing the model because you have everything inside one system. You don't need to switch anything or break any automations or anything like that.

So if you want to get all of our best training on this sort of stuff and our agentic operating system ready to go, you can get that inside the AI Profit Boarding, link in the comments description, or just go to the AIProfitBoarding.com. Inside the community, you can ask questions that I personally answer them. Plus there's a community online 24 seven ready to help whenever you need to. And then inside the classroom as well, you can get our new daily updates over here.

So if you go to the AgentOS system, you can grab the AgentOS over here with the zip file, Quen 3.8 plugged in, the video tutorial, and then we have all of these new tutorials based on what's actually useful and working. And then inside the calendar, you can jump weekly coaching calls, share your screen, ask questions inside the map. You can meet people in your local area who are building out with AI agents like you and inside the community, you can get access to an amazing network of people. So feel free to get this, link in the comments description, or go to the AIProfitBoarding.com.