HOLY CRAP. Bloomberg has the goods on Gemini 4! "it does less well when employees actually put it to work .. The model struggles to handle certain coding tasks" Bloomberg: "As Alphabet Inc.’s Google prepares for the coming launch of Gemini 4, it’s grappling with internal skepticism over how well the flagship artificial intelligence model performs in key areas, such as coding. While Gemini 4 has performed well on benchmarks the industry uses to gauge model efficacy, it does less well when employees actually put it to work, according to people with direct access to the effort. The model struggles to handle certain coding tasks, said the people, who requested anonymity to discuss an internal matter."

Claude Opus 5.5 has the best visual design of any model I have tested so far

Both models regress in GDPval-AA v2.1 at max effort, driven by shorter deliverables that more often omit required elements.

iMessage pro tips: ⌘+e edits your last-sent message ⌘+r replies to the latest message ⌘+t then 1-6, to tapback the latest message

Mark, by @colossusjeremy. This is a story about the head of history's largest nonterritorial empire, who nonetheless remains a prisoner of his rivals. Has he finally found a way out? The profile has been nearly a year in the making. Stern interviewed three dozen people, including Zuckerberg, his MMA coach, and his critics and competitors at the frontier AI labs. He spoke with his parents, Karen and Ed, and watched their son get punched in the face. The story spans Meta HQ in Menlo Park, the Zuckerbergs' homes in Palo Alto and Lake Tahoe, and a party in a Manhattan warehouse with Kylie Jenner and Timothée Chalamet. There is also a sneak peek of Aaron Sorkin's forthcoming sequel about Zuckerberg, sourced via a Hollywood fixer Stern first encountered in Russia. The result is a profile unlike one you've read before. colossus.com/article/mark-zu…