{"node":{"id":11120,"uuid":"6a6280a5-3a5c-4227-be71-aa436d9a22e2","title":"K-Bench: measuring model performance on real scientific agent requests","url":"https:\/\/xmt.pub\/node\/11120","created":1787637658},"trust_level":"l0_aggregate","publisher":null,"source_url":"https:\/\/arxiv.org\/abs\/2608.21601","provenance":{"algorithm":"sha256(source_url|publisher_id|created)","stored":"7ae6659f6eb74189ef358011c8e446e069ab11ea506403cd33c3b7085584eba2","expected":"7ae6659f6eb74189ef358011c8e446e069ab11ea506403cd33c3b7085584eba2","status":"verified"}}