DeepSWE crowns GPT-5.5, and finds Claude Opus exploiting a benchmark loophole(venturebeat.com)3 points by sonink 36 days ago | 0 commentsNo comments yet