We are open-sourcing V1.0 of our agentic robotics framework, ASPIRE, with full real robot skill discovery and self improvement capabilities! On the day Astra arrived, our lab switched the backend model and saturated all the benchmarks to ~99% over night. And the frontier is already on the move: - Opus 5.5 slightly edges out Astra on our robot simulation bench. - Sonnet 5.5 hits the Pareto frontier of cost vs. robot capability, $12/task. We first introduced ASPIRE and /robot skill library in June, way before the big models could reliably do robotics. A rising frontier tide lifts all boats, but Astra was a 0->1 tsunami on many tasks. Now you can also feel the Physical Bitter Lesson at home like we did on that fateful day ;) links in thread

JUST IN: As robots move out of cages and into shared spaces with people, how do we actually test whether they’re safe? That’s the problem @safeworldai is solving. The company just emerged from stealth with $12M led by @shinecapital and @speedrun. I spoke with co-founder @kwong47 about what they’re building, and the way he described it stuck with me: “Think of it like CI/CD for robots.” Every time a robot’s software or AI model changes, Safeworld runs it through simulation-based safety tests automatically. The same way software teams run test suites before shipping code, robot teams can now run safety checks before deploying updates. Safeworld injects e.g. simulated humans into robot log data to test how the robot detects people in rare, dangerous scenarios: blind corners, workers tripping and falling, humans crouching or running, edge cases you simply can’t test at scale in the physical world. The focus is mobile manipulators specifically. Wheeled and legged humanoids, operating outside cages, in real environments alongside real people. And the founding team has the right pedigree for this. Co-founder Ding Zhao directs the Safe AI Lab at Carnegie Mellon, he’s spent nearly his entire career on exactly this problem. The industry is moving fast to deploy generalist robots at scale. The safety infrastructure to validate them has to keep up! ~~  ♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com

In April 2013, a 58-year-old mathematician named Yitang Zhang submitted a paper to the prestigious Annals of Mathematics. At the time, he was virtually unknown in the mathematical community. After earning his PhD from Purdue University in 1991, Zhang struggled to find a permanent academic position. For years, he worked outside academia, including at a Subway sandwich shop, before eventually becoming a lecturer at the University of New Hampshire, where he taught undergraduate mathematics. Yet throughout those difficult years, he continued pursuing his own mathematical research, particularly the mysteries of prime numbers. In July 2012, while visiting a friend in Colorado, Zhang was sitting quietly in the backyard when an idea came to him. He had spent three years working on a problem involving the gaps between prime numbers, and suddenly he saw a way past an obstacle that had frustrated some of the world's leading number theorists. Over the following months, he carefully developed that insight into a rigorous mathematical proof. On April 17, 2013, he submitted his manuscript to the Annals of Mathematics. The response was remarkable. Within just three weeks, the referees had delivered an unusually enthusiastic report, recognizing the work as a landmark achievement. The paper was formally accepted on May 21. For centuries, mathematicians had wondered whether infinitely many pairs of prime numbers could occur within some fixed distance of one another. The famous twin prime conjecture predicts that infinitely many pairs differ by exactly 2, such as 11 and 13 or 17 and 19. Zhang did not prove the twin prime conjecture itself. Instead, he established something that had never been proved before: there are infinitely many pairs of consecutive primes whose difference is less than 70 million. Seventy million might seem enormous when compared with 2. But the crucial breakthrough was that the gap was bounded by a fixed number, no matter how far along the number line one went. It was the first proof that bounded gaps between consecutive primes occur infinitely often. His discovery opened an entirely new chapter in number theory, inspiring mathematicians around the world to improve his bound. In 2014, Zhang received a MacArthur Fellowship in recognition of his extraordinary achievement.

Introducing The Cheeky Pint Benchmark. We made our intern watch every episode of Cheeky Pint to bring you this breakdown of pints drank by guests to the nearest sixth. Congratulations to @elonmusk who is currently in the lead with three full pints of Guinness.

I wish to present my new favorite square packing problem: 29 squares. Not for the (currently best known) packing itself; it's a bit messier than 17, but not unexpectedly so. But rather for the side length, which is truly cursed.