Hacker News new | ask | show | jobs
DeepSWE crowns GPT-5.5, and finds Claude Opus exploiting a benchmark loophole (venturebeat.com)
3 points by sonink 18 days ago