
ARC-AGI-3 Just Dropped the Hardest Reasoning Benchmark Yet. Frontier Models Are Already Failing It.
Thirtythree percent. That's the best score any frontier model has managed on ARCAGI2 after months of optimization, promp...
Full-stack developer and open-source advocate. 15 years in the industry.

Thirtythree percent. That's the best score any frontier model has managed on ARCAGI2 after months of optimization, promp...

Twelve people in a courtroom just did what thousands of engineers, product managers, and policy teams spent a decade avo...

417 points on Hacker News. Not for a new framework, not for a benchmark, not for some hot take about AI replacing develo...

268 kilobytes. That's what Video.js weighed in 2024, right before its original creator, Steve Heffernan, took the projec...

Four million monthly downloads. Over 100 LLM providers supported. One compromised package sitting between your applicati...

Last Tuesday, Gerd Faltings won the Abel Prize for proving the Mordell conjecture , the kind of work that reshapes a fi...

Someone at Microsoft finally said it out loud. Scott Hanselman , one of the company's most visible developer advocates,...