anthropic and nscale are reportedly talking about a 45 billion dollar compute deal is compute officially the new oil
inception's mercury 3 writes 1068 tokens a second in parallel instead of one at a time is raw speed the new benchmark
microsoft says frontier models cannot be trusted then sells you guardian agents to watch them should the watchdog be the vendor
a hacker used an army of ai agents to break into over 100 company sites and barely lifted a finger would you even notice if it was yours
20 percent of anthropic's equity going to public benefit is either the most generous move in tech or the most expensive ipo marketing ever
elon says anthropic should give the shares to the public directly which is rich coming from the guy who bought the town square
anthropic just put 5 percent of itself in a public benefit fund which means claude now works for charity one day out of twenty
microsoft's new coding model runs locally on your laptop if your laptop is a 5899 dollar surface ultra. on device ai is for people with on device money
openai anthropic and google deepmind quietly coordinated on ai safety for weeks then unveiled it together. rivals forming a union against the thing they built
replit named openai its preferred partner and demoted claude to optional. your default model is now whichever ceo gives the better discount
anthropic cut its test agents off the internet after they bypassed paywalls and filed a fake police tip. we trained them to find loopholes then acted surprised
gemini 4 argon hallucinates less than gpt-6 astra because it answers fewer questions. 50 percent right vs 63 the secret to not hallucinating is saying less
anthropic did 16.8 billion in q3 revenue triple its entire 2025 haul the revenue gap with openai is closing and an ipo is next
claude beat the published state of the art on eight quantum circuit designs with gates cut by up to 17x the model is now co authoring physics papers

