Claude Fable 5.2 beat protein-only AI at its own game in a new test
A test of 22 language models finds only the newest Claude models predict little-studied proteins, and Fable 5.2 beats the open protein models tested.
A test of 22 language models finds only the newest Claude models predict little-studied proteins, and Fable 5.2 beats the open protein models tested.
Palo Alto Networks will pay $4.5 billion in stock for Applied Compute, a startup that tailors open-weight models to company data.
A higher price, a bigger offering and $26 billion pledged by chip and cloud companies: the prospectus spells out Anthropic’s Nasdaq debut.
Microsoft’s To Do apps shut down on June 30, 2027. Work accounts move to Planner without their tasks, and personal accounts get no replacement.
About 50 large sites take nearly all of that power. In the EU in 2024, only Ireland’s data centers took a bigger share.
OpenAI says its models found up to 14% gains on AMD’s next GPU. Anthropic says Claude ported a model to AMD in days.
In 16 countries, a median of 33% of 15- to 22-year-olds have used AI as a friend: 46% in Australia, 16% in Kenya.
Cloudflare released the weights of Clairic and Clairic-flash, two Apache 2.0 models that extract data from documents, with scores it measured itself.
Google’s fourth Gemini 4 model in seven weeks reaches Pro subscribers. It ranks high, but its safety report is still pending.
One setting for history, another for AI training: Google is updating Gemini’s controls and will keep History-off chats for around two months.