China's Qwen 3.8 Max DESTROYS Kimi K3?

We got another big launch today. Quen 3.8 second only to Fable 5, according to Quen's announcement. Here's open source going live soon. It's not available to code with unless you're in mainland China.

I did check out the international website as you can see, but Quen 3.8 is not available on there just yet. However, what I will say is this is a big one, a big one, my friends, because. We have a 2.4 trillion parameter model, which is absolutely massive. The model is continuously evolving.

They believe according to Quen and just test this stuff out for yourself. Don't believe everything you read as always, but they believe it's one of the most powerful models available today. Compatible to leading frontier AI models second only to Fable 5. Now that is a huge statement because number one, it leapfrogs GPT 5.6 and number two, this has already, according to them, beaten Kimi K3.

This was an announcement just made a few hours ago. So literally about one hour and 10 minutes ago and pretty amazing stuff to see this. Now, Chinese AI is controversial. I mean, it's, it is overtaking very quickly stuff like for example, um, GPT 5.6 Fable 5 is kind of only available on 50% of the usage plan of Claude, which means like you don't really properly have access to it and these open source models, uh, three.

So for example, like, I mean, you're not going to host them yourself. Like they're too big to host yourself. You need a ridiculous amount of DGX sparks connected together. But if you look at probably the leading model in the world right now, which Kimi K3 actually beat on my tests.

Um, a lot of models, for example, like Fable 5 and GPT 5.6. So on the benchmarks, these models win, but on the actual tests that I do personally on GoldieBench, which you can check out on goldiebench.com. Kimi is crushing them. So if this comes out and it's second only to Kuen 3.8, we are looking at a very strange world where two Chinese models have leapfrogged chat GPT in the space of a week, bear in mind.

There doesn't seem to be like any preview. Um, you know, preview sort of system where, for example, if you look at GPT 5.6, you couldn't access it for two weeks, it was only available to like cybersecurity firms. If you look at, for example, Fable 5 comes out for three days, get shut down, comes out again, never really feels quite the same. And you can only access it on a limited basis, a very limited basis.

And we don't know if it's going to be on the core plan forever. However, you look at Chinese models, open source models, like for example, Kuen 3.8, like for example, uh, Kimi, they're available for free. They're going open source. You can access them inside the chat at Kimi.com for free.

And also they perform very, very well. In fact, to the point where, for example, if you look at Kimi K3, it's actually outperforming on coding benchmarks, other models, like for example, Fable 5. It's also top of the front end, uh, coding benchmarks. So you can see it is crushing, uh, Fable 5 and GPT 5.6.

It's just a weird week. Like this is, everything is changing. I feel like the table is just showing every day. And we are looking at a situation here where the Chinese models are just getting ahead of the U S models.

And there's going to be more and more sort of restrictions on open source models, potentially in the future. This actually brings me on to something that was looking at from Dean W Ball, head of strategic futures at OpenAI. And he did like quite a really interesting post here. Um, obviously like GPT 5.6 or the OpenAI team, they are worried about Chinese models.

You've got Fable 5, who I'm sure are worried about Chinese models as well. Um, there's also been rumors of like distillation as well. Um, however, what Dean was saying on the latest update from Kimi K3 this week is that it's a very good model. However, he don't think, and he doesn't think that it's performance is explained by distillation.

So distillation is kind of like where you test the model thousands of times, and then you understand the patterns and then you train your own AI model to think in that way. And they're very simplified explanation. That's how it works, right? I'm sure it's much more complicated in reality, but basically what they're saying now is that Kimi K3 seems pretty much on par with the best public models.

This is the head of futures at OpenAI, so it's not like some random Gita on Twitter saying this. And so he also said it's very token hungry, um, and not that cheap to run. But actually if you just use it on the coding plan, it is cheap to run because then do you have your set limit? You're not paying per API.

Now also, he's also said that he's personally surprised the Chinese state continues to allow the open sourcing of models this good, given the potential risks. However, like if you look at the OpenAI team, they number one, have a lack of, uh, inference, right? Because for example, like there's all these export controls on chips going to China, which means like if they open source it, then people just hosted themselves or you get, for example, Olama and companies like that hosting Kimi and then giving it away to people. Um, which means like these open source models can be distributed more without China actually having the chips to deploy them and to make it publicly available as much, then you also got an interesting situation here where basically he said the one probable outcome of an open weight model, dominant world, AKA, you know, and this is just, this was before Quen 3.8 was announced, but now after Quen 3.8 has been announced today, this goes even more wild.

Does that make sense? Because he said basically one probable outcome of an open weight model, dominant world is cool AI communism. Um, which is an interesting one where basically AI is public good, which is ultimately provided by the state as a kind of digital public infrastructure. Um, and he says this future strikes me as very dystopian, but I mean, it's one of those, isn't it where like, do you want the models restricted or do you want the models open source for everyone?

Or how do you want to do this? Because like, for example, GPT 5.6, it was restricted. We couldn't access it anyway. Now it's out and we can access it, but actually there's a lot of models out there that we can't even access.

And then you've got, for example, like Fable 5, it's like, well, you can't really access that properly. It was taken down and then brought back out by the government. It's like, I don't know. I mean, it's just an interesting world to see.

Okay. These are some really interesting reflections on open source. And it's the first time I've seen public figures from companies like OpenAI actually take open source models seriously. First time in history, to be honest.

And then also you said, uh, I would guess the Trump administration will at some point realize the best strategy here is to create large amounts of regulatory risk around the use of open weight models, which, yeah, I can kind of see that, so like basically they would say, don't use public models, or for example, don't use open weight models. Or if you're like an enterprise firm in the US, maybe it just gets banned, something like that. And then the growth of open source models is either like smeared by the US government, if they don't want it to grow so much, if they don't want China to sort of take over, or they could actually ban it, you know, kind of like they did with Claude where they said with Claude that it was, I think they put export controls on Claude basically. So like, if you're working with the government, you couldn't use Claude inside your company.

That was one of the situations that happened a couple of months ago when things weren't working out so well between the US government and Claude. But yeah, it's all getting a bit political, mate, to be honest with you. Um, let's have a look here. It's probably true that open weight models of this capability make the world a bit more dangerous, according to Dean.

It's an interesting one, isn't it? It is an interesting one. I don't know, but it'd be interesting to know, like if you're watching this, what do you think about Chinese models? How do you feel about the whole situation?

How do you feel about the US falling behind China? These are all questions that need answering, you know, as someone from the UK, we haven't even got anyone in the race, to be honest with you. And also these models are extremely impressive. I mean, for example, if we go over to GoldieBench here, if you look at some of the stuff that we created with KimiK3, it's outrageously good.

Um, it performs very well on pretty much everything. It can create beautiful stuff like this. And also it was pretty fast when we were using it. Like we built out about 50 tests in the space of 24 hours.

The things that you can create with it are incredible, but not just that. The thing that actually blows me away about all of this, right, is that Chinese models and open source models have become so much better, so much faster. It's almost like you've seen Chakchipiti and Claude, who were miles ahead in the race, get leapfrogged overnight. So who knows what's going to happen with Quen 3.8.

I'm excited to try it when it does come out. I would expect this to come out within the next 24 hours, but let's see what happens next. It's not on open router yet from what I checked earlier, but I'd imagine that will be dropped soon because for example, Quen 3.7 was dropped pretty quickly on open router. Um, as soon as it came out.

So it's not available just yet with Quen 3.8, but I'd watch out for that. And yeah, very interesting times peeps. So thanks for watching. If you do want to get more trainings on, for example, KimiK3 or my best trainings on AI automation, how to use it, et cetera, you can get that inside the AI Profitable Boarding.

Link in the comments description or just go to the AI Profitable Boarding dot com. Inside the community, you can post questions, get help and support whenever you need to. Inside the classroom, you can jump on our coaching calls and you can also get access to all of our best trainings. We have lots of great courses, step-by-step.

So we actually have a full one hour course on KimiK3 and GPT 5.6 and Codex and Grok and how to build an agent operating system and everything else. Inside the new daily update section here, you can get all of our best tutorials and all the amazing stuff that's been dropping and we add new. Tutorials based on what's working inside the map. You can meet people in your local area.

Who are building with AI agents like you and this all inside the AI Profitable Boarding. Link in the comments description or go to the AI Profitable Boarding dot com. Thanks for watching. I'll keep you posted.

And when Quen 3.8 max is available to everyone, I will be testing out. So I can't wait to try that. Cheers.