Hot take: OpenAI is holding GPT-6.2 Astra until just before Anthropic’s IPO because OpenAI thinks Astra can beat Claude, so that’s the moment to steal the spotlight. Ant IPO is planned for November 19th.

The AI Futures Project now forecasts that AI capable of matching top human AI researchers will likely arrive in 2027, absent a major slowdown. We are 5 months ahead of the curve. The update follows METR’s estimate that Anthropic’s AI progress has roughly doubled in speed, with AI contributing much of the acceleration. We are even outpacing expected exponentials. 2027 will be the most crazy year of humanity.

1/ I had early access to Google's new Gemini 4 Carbon in Antigravity. I gave it 13 coding briefs, from 3D to websites and apps. Here's what came back. Its very impressive. First: ORRERY. One control drives the gears and folding petals, even in reverse. One follow-up improved the framing. Several more examples and my resume. A 🧵:

1/ I had early access to Google's new Gemini 4 Carbon in Antigravity. I gave it 13 coding briefs, from 3D to websites and apps. Here's what came back. Its very impressive. First: ORRERY. One control drives the gears and folding petals, even in reverse. One follow-up improved the framing. Several more examples and my resume. A 🧵:

Robots need examples of the small corrections people make without thinking: adjusting a grip, shifting a part, trying again. Memories Upload labels actions in first-person video. More varied examples could help robots handle work beyond neatly staged demos.

Super impressive: Google DeepMind has unveiled Gemini Robotics 2.5, an AI model tested on more than 200 robot tasks, including tying bows and handling delicate objects. Robotic is the next upcoming revolution.

This is such an important win for Fable 5.2. It makes by far the most significant leap in one of the most important benchmarks that measures real-world work. On the Remote Labor Index, it completed 41.3% of projects to at least the standard of the human original. Fable 5.1 scored 17.9%. That’s more than twice as many projects successfully completed. The test gives AI agents real briefs and files from paid freelance projects, including software development, design and data analysis. Human reviewers compare the results with the original work. So yeah, AI comes for real labor now. First slowly, suddenly all at once. Very important benchmark, very significant jump.

AI might be the biggest lever kids have ever had for learning. Every child can now have a patient conversation partner that's there whenever they get curious. The catch is that most AI products are built to remove steps. In learning, the steps are where understanding comes from. @gakkuai is built around that. It's an AI learning app for kids aged 6 to 16. Its companion Raku asks the questions and pushes back, and the child does the explaining and writing. In this conversation Raku interviews co-founder @anniechen419. What's interesting is her take on prompting. A good prompt starts with knowing what you want and which context matters, so it's really an exercise in clear thinking. Kids who learn to explain their ideas are practicing the same skill. That's the version of AI in education I find most exciting.

Official: Anthropic has launched the roadshow for its proposed IPO, offering 825 million shares at $70 to $78 each. That implies an estimated fully diluted valuation of roughly $1.24 trillion to $1.38 trillion, below earlier expectations of around $1.5 trillion.

While we’re all focused on Claude Fable 5.2’s coding abilities, its work in science and engineering has received far less attention. Anthropic has made insane steps and solved crazy hard problems that many missed: - Hardware: Fable 5.2 designed three electronic devices from short prompts, including their circuit boards, enclosures and software. Partner Diode built them, and all three worked without changes to their circuitry. - Gene therapy research: Mythos 5.2 developed a method to model the complex protein shells of viruses used to deliver gene therapies. In Anthropic’s test, it accurately modeled 28 out of 34. - Chemistry: An early version of Mythos 5.2 planned ways to synthesize complex molecules. Three chemists rated the work around the level of an early PhD student. - Physics: Anthropic says Fable 5.2 solved two open theoretical physics problems involving black holes. The results have been published as preprints. - Materials science: Discovered Materials says Fable 5.2 made a calculation used to check material stability 36 times faster on average. Across 13 test cases, the total time fell from 3.3 hours to under four minutes. - Robotics: In Anthropic’s simulated RoboDojo test, Fable 5.2 achieved a 64.9% average success rate, up from 27.6% for Fable 5.1. So yeah, this is nuts. And at the same time thats why we need to accelerate. AI helps us solve problems in *ever* scientific field. And yet, we are still in 2026, and this is still the beginning. Fable 5.2 shows us where we are heading. And 2027 is going to nuts.

Fable 5.2 is the best model I've ever used. It finds bugs and fixes them that Opus 5.5 just missed. Super impressive, insane model. But oh boy does it eat through your rates fast, especially 5h limit. Even on medium effort.

As expected, Anthropic has a significantly better model ready to deploy internally, one they aren’t sharing with us peasants. We have to make do with Fable 5.2, as they themselves indicate in their own paper.

It seems that internally, Google is either truly heading toward RSI or has already reached it. I can’t explain it any other way. Currently, numerous leaks are circulating regarding "Krypton" - Google’s latest model - which is truly breathtaking. Google is back in a massive way. I cannot verify this myself, but I’ve linked a few impressive outputs below:

I can confirm that GPT-6.1 Sol has already routed me to the more current model (GPT-6.2 Sol?). Two testings: -(Simpson) 742 Evergreen Terrace as an interactive Three.js diorama -Pelican

US professional and business services added 70,000 jobs in October, despite fears of widespread AI-driven job losses. But some sectors are shrinking: employment in document preparation fell 17% year over year through September. via WSJ IMHO This is because the adoption of AI is only just beginning to take hold, and agentic AI has not yet truly made an impact in the professional world. We are, therefore, only seeing the early precursors.

Meta is joining the race for scientific breakthroughs using AI. As a result, all major frontier labs except SpaceXSI are now intensively working to advance scientific discovery. The competition will lead to unexpected breakthroughs here.

I'm hearing more and more rumors that Qwen 4 was still dropped today. It's supposedly expected to be a positive surprise. The CTO recently said that Qwen 4 was dropped this week, so it should happen today. Fingers crossed.