THIS WEEK IN AI: Grok is Back, ChatGPT 5.6, Meta's New Models, China's Threats

**Josh** (0:00)
I've been waiting so long to say this.

**Ejaaz** (0:02)
Elon is officially back. The first major AI model released from SpaceX AI since they went public, since the $60 billion cursor acquisition is officially live. It's called Grok 4.5, and it's pretty damn good. It ranks around 3 after Fable 5 and Gpt 5.5, but it's really good at coding, and it's astonishingly cheap. It's around half the cost of Opus 4.8, and it's their quickest model yet. But you don't only got to enjoy this for about a day or maybe even a couple of hours, because Gpt 5.6 was officially released from OpenAI. And if you're listening to this, you probably have access to it right now, a general release. It's their best model yet, extremely good at agentic reasoning, but not quite Fable worthy. But they are training Gpt 6, which is going to be released supposedly a month from now. If you're releasing models, you're probably thinking that's amazing for the AI world. But this comes right at a time where our main adversary in China has been releasing so many models that are open source and everyone gets to use it. They're super cheap, but they might be considering cutting off the West from using their models. Meanwhile, Meta, who spent around $25 billion trying to build the best model and have seemingly failed, just released two new models, their Metamuse Image and Metamuse Video Models. We're going to get into all of this and more on today's episode.

**SPEAKER_3** (1:19)
There's a lot going on. It feels like nothing happens for a week or two and five different releases happened all on one day. That's what just happened. We got new JAT GPT. We have a new voice interface, which I love that we'll get into. But I think we do have to start with Grok because, like you said, Elon is back. The SpaceX AI team, which is now officially formalized, they have a logo, they have their official Twitter handle on X. This is their first release in a long time and the results are pretty remarkable. They're classifying this as Opus class on a relative basis compared to Cloud Opus 4.8.
It seems to have actually accomplished that on all of the benchmarks. In fact, it even exceeds it slightly in some benchmarks for a fraction of the cost. I think this is one of the more interesting things about this model is just how efficient it is for how much intelligence you get out. I know that the XAI team or the SpaceX AI team, they've always been really insistent on maximizing the amount of intelligence per watt of compute. It seems like this is the first sign of that actually working. For the same amount of power and electricity, they're able to generate tokens that have more intelligence per token than just about anyone else. Like you mentioned, Ejaaz, the new Grok model is being trained. It is currently cooking in the oven. They have all the GPUs humming. And when that comes out, I really believe that they're going to have a shot at the top. And we talked about this way in the past, maybe like two or three months ago, about how the SpaceX AI opportunities really just in their ability to build out infrastructure quickly. And as a result, they actually had too much infrastructure for their use cases. They were selling it off to Amazon. They were selling it off to Anthropic because they just had access and created this huge business model for them. But now they actually have a use case for this. They're starting to take these GPUs, they're starting to train the models, and the results are actually great. And it's like, okay, wow, they were required for a couple of months, but now here they are, right back on the frontier. And you have to imagine that the next model that's being trained on however many trillions of parameters is going to be right up there, at least very close to the frontier, for what I would imagine is a very, very competitive price.

**Josh** (3:13)
I think the most hilarious part about this is this is probably Elon's most humble model launch ever.

**Ejaaz** (3:18)
He was super up front saying this is Opus 4.8 class grade model. It's not as good as Fable, but we are training a much larger model. Speaking of size, the size of this model, Grok 4.5, is only around 1.4 trillion parameter models. Now, if you compare that to Fable 5, that's around 10 to 15 trillion parameters in size. So the fact that they were able to achieve number three on some of these coding benchmarks, like the one that we're showing on your screen right now, DeepSui Bench, is pretty impressive. You can only imagine what a model maybe five times the size of this could probably be quite competitive. And this brings me to the absolute Hail Mary move that Elon and SpaceX AI pulled off a few months ago, or actually less than a month ago, which was the $60 billion acquisition of Cursor. You're wondering. Yeah, it paid off massively.