Jev matched a thinking model on a real data-interpretation task at 1% error, in seconds instead of minutes, but only after the questions were redesigned; the questions and the code did most of the work.
Typed, calibrated, parallel decisions at 100 milliseconds are a more natural unit of enterprise AI than generated prose, and they collapse the cost of intelligence inside a decision process by orders of magnitude.
By treating frontier models as components to buy, benchmark and replace, Apple is betting its AI advantage will come from silicon, privacy and the integration layer.
Promise 4x, charge 2x, get ~1.5x — not a great way to get trust.
Anyone trying to get something done inside a company faces an implicit choice: how much to settle before they start.
Anyone using AI – specifically, LLMs – eventually has to deal with the same question.
Cursor's maker Anysphere announced a $2.3b Series D in November 2025 - $29.3b post-money valuation, past a billion in annualised revenue and accelerating. Bloomberg reported it topped $2b by February 2026.
Yesterday I posted about TurboQuant, a Google Research paper that compresses the main memory bottleneck in large language models by over 6x.