**Julian Goldie** (0:00)
Grok 4.5 has been announced. Now, this is a brand new update coming from Grok, and it's based on a 1.5 trillion V9 foundation model. You can see the announcement from Elon Musk right here. With Cursor data, bear in mind that Cursor have just been acquired by Grok, and that adds in additional data for supplemental training, and it's now in private beta at SpaceX and Tesla, which means that it should be dropping soon. I'll be talking about pipelines in a second. If you're wondering, okay, how does this look? We're looking at early evaluations showing performance close to perhaps exceeding Opus 4.8. Now, something to bear in mind here.
It's only been recently, and I've done another tutorial on this, that I've been impressed by Grok. So previous announcements didn't really excite me, but at this point, when I've tested Grok, and I'll show you what we've built with actually inside the AI Agent Operating System, it created some awesome stuff. So if this is a step up, and if you can use it with your existing Twitter subscription, wow, plus bear in mind, like with Opus, you can't even use a wharf inside Hermes or your AI agents. What that means essentially is that if you're building with this stuff, well, good luck, mate, because if you're building with Claude Opus, you have to use the API inside your AI agents, and that gets expensive. Whereas, for example, if you can build with Grok using your AI agents like Hermes or OpenClaw using the ORF and you're existing Twitter subscription, that for me personally wouldn't cost me an extra money. That is fantastic. So it's going to be exciting to see where we go with this. Let me show you some other stuff that we built with this as well. It's a pretty good, pretty powerful model. I mean, we built all of this using Grok Build, and this is a full open world game that are basically mocked up. And if we actually check out Goldie Bench, Grok Build was way up there. And I say that not to like get over hyped or over excited. But if we look at some of the stuff we built with it, it's pretty good. It's pretty good. So having said that, where are we up to with Grok 4.5? So it's in early beta. And we'll talk about timelines. I've actually seen some recent timeline drops and how that compares to when the announcements come out. So typically when this is announced in beta, there's a typical timeline related to that, which is when it will drop to the public. We'll talk about that in a second.
So what you can see here is that Elon has clarified, to be clear, I'm not saying the Grok V9 foundation model will be mind-blowingly better than anything, but it will be a solid workhorse in the same league as Opus. And again, like if you already subscribed to Twitter, well, you don't need to pay an extra if you are using an API. Well, the great thing about this is you can use a wolf inside Hermes instead of having to pay for a Claude Opus API. So it's pretty powerful in that way. And bear in mind, like it is getting dramatically better. It is improving all the time from what I've seen from Grok Build. So I'm excited to see what this is. Now, some people are going to say it's the Opus killer. And there's a big cap between some of the hype here. So just be careful of that too. So some people have said, you know, this is going to take over Opus.
Polymarket gives it a 42% chance of its release in July, by the way. So if you're wondering when the timelines are, that's one estimation. Make your own mind up, you know, believe what you want.
That sounds about right to me, to be fair. But the fact that Grok 4.5 is already trained and actively being used by his own companies is a pretty good sign. And if you're wondering, okay, when do the normal sort of things come out?
This is a typical timeline of the announcements. So if you look recently at the previous stuff, you've got, for example, Grok 4.3, Grok 4.4, late May. So there was quite a long gap between those two. And then you got Grok 4.5, V9. And then we've got the Grok 5 Target, which apparently, according to some rumors, is going to be a 10 trillion model, which is pretty insane. I mean, I don't know if anyone saw the announcements about Le Chaton Flat, but that kind of sounds in that category. So again, be skeptical with this stuff. Make your own mind up. The main thing that I would say is like, this is exciting. I'm excited to see what happens next. It should be a pretty good model, to be fair, and it's only getting better. And if it overtakes something like Opus 4.8, that wouldn't surprise me because, you know, if you look at the way Anthropic are going, like they're just getting gated.
9 more minutes of transcript below
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/1000774913014