GLM-5.2 (max) matches Claude Opus 4.8 on Harvey LAB-AA benchmark | Dark Hacker News