Qwen 3.8 Max Challenges Kimi K3 artwork

Qwen 3.8 Max Challenges Kimi K3

The Daily AI Show

July 20, 2026

The episode opened with the impact of Kimi K3 and Alibaba’s new Qwen 3.8 Max model. The hosts discussed whether the latest Chinese open weight models are now reaching or passing frontier-level coding performance, while also warning that early benchmark claims still need real-world validation.
Speakers: Brian, Andy, Beth
**Brian** (0:00)
Hey, what's going on, everybody? It's Monday, July 20th, 2026 And this is The Daily AI Show live. We're live at 10 a.m. Eastern and 7 a.m. on Pacific and then all around the world, obviously. With me today, Andy and Beth and I'm Brian. Hope everybody had a great weekend. I was, you know, somewhat sad to see the World Cup come to an end. It's been a lot of fun. And so the the final was yesterday. Should you be following those types of sports or whatever? But it was it was a lot of fun. I'm really I'm sad to see it go, but I think more so because of it being hosted in North America and seeing all the real success and as most people have been talking about hearing foreigners from all over the world literally come to the United States and Canada and Mexico, but mostly the US and just talk about how it was different than whatever they expected when they showed up in a positive way. So sad to see it end in that way, but that doesn't stop the AI news train from continuing on.
We got all sorts of stuff going on. We got the head of strategic futures at OpenAI stepping on their own foot a little bit and then trying to course correct. We got model updates for Kimi K3, as well as what is Anthropic finally doing with Fable. Do we have Fable today? Do we not have Fable today? We got to talk about that. And yeah, all sorts of other stuff. I have stuff from the news and from the movies and everything else.
But with that, I'll kick it over to Beth or Andy. To kind of start off, what caught your eye over the weekend?

**Andy** (1:33)
Well, there's lots of discussion about the impact that Kimi K3 is having. Because it was released last week and has thrown a big set of waves into the AI pond. But on Sunday, Alibaba released their latest Qwen model 3.8 Max. And it is like Kimi K3. It's a multi-trillion parameter model. It has 2.4 trillion parameters, just shy of the 2.8 trillion parameters of Kimi K3. But it beats everyone now on coding.
So it's available on Coder as a preview. And as a result of the performance on that open arena kind of testing, that is the first early signals of just how good a new model is.
Alibaba Group's stock rose 5.4 percent today, Monday, as a result of that release. So now there's not just one major Chinese open. And by the way, they're saying it's going to go open weight as well. It'll take a little while for them to publish all the open weights and so on. But now you have two Chinese models at frontier level and actually surpassing Fable and Sol in some domains.

**Brian** (3:07)
Well, and I think some of this needs to be with a grain of salt because a lot of it is every report, you read a Forbes, you whatever, they will usually throw the word unverified in there somewhere, that particular word, Kimi as well.
I think there's a bit of wait and see. For instance, with Kimi K3, people have said it's a bit token hungry and so that obviously is going to impact cost.
On that level, this is from kimi.ai. Kimi K3 has received far more love, this is on X by the way, than we expected and our GPUs are feeling it. Over the past 48 hours, the band has pushed close to the limits of our current capacity to protect the experience of existing subscribers. We're temporarily pausing new subscriptions and prioritizing compute for our current members. As well, we'll also split membership into two more focused plans, the Kimi membership for KimiWeb app and work and KimiCode membership for coding workflows. This will help us match compute more precisely and keep the experience stable. And then we thank you for your patience and understanding. At the same time that was happening, Anthropic, will they, won't they?
A saga that's been going on, said beginning July 20th, that would be today, Claude Fable 5 will be included in all Max and Team premium plans at 50 percent of limits. So Pro and Team standard users will continue to access the Fable via usage credits and will receive a one-time $100 credit. I have a story about that, my $100 credit. Demand for Fable has been challenging to predict, which is why we're rolling it out. We rolled it out to subscription plans and stages, extending access several times as we secured additional capacity. So this is all like, to your point, Andy, we had heard there was rumors that Alibaba was probably going to have an answer to Moonshot, Kimi K3.

52 more minutes of transcript below

Feed this to your agent

Try it now — copy, paste, done:

curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire.

Using your own key:

curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000777613466