Opus 4.5 Steamrolls the Competition
PLUS: Schools brace for AI homework takeover
Karpathy tells school boards AI homework detectors are doomed, so flip to in-class testing instead.
Anthropic's new Opus 4.5 just became the best coding model in existence, full stop.
Revolut hits a $75 billion valuation, still Europe's most valuable private company.
Twitter's launching a hiring feature and people think LinkedIn is toast.
Opus 4.5 crushed every human candidate ever on Anthropic's hardest engineering exam.
One engineer defends lowercase writing as cleaner and more elegant than arbitrary capitalization rules.
ChatGPT is losing ground as smarter models like Claude pull users away. Product scaffolding can't save you.
Marc Andreessen says writing things down gives you power because so few people actually do it.
V0 confirms Opus 4.5 is unreasonably good at Next.js, their best model yet.
Cursor still dominates after Google's antigravity launch failed to dethrone it.
Opus 4.5 hit 80.9% on SWE-Bench, the wildest benchmark in the AI race.
Amazon Leo Ultra delivers 1 Gbps satellite internet with custom silicon, bypassing public internet entirely.
Opus 4.5 does better without extended reasoning tokens than with them. Super token efficient.
Gemini 3's launch spiked market share from 23% to 30% overnight.
The work is shifting from producing knowledge to designing processes that generate it for you.
Chris Sacca says lean into your disproportionate reactions, that's where the interesting stuff lives.
Wispr ditched ARR as their north star because millions in ARR means nothing if users churn fast.