OpinionThe Editorial Board
A model’s label is no substitute for independent results
Artificial Analysis scored the preview 38, behind five Chinese open models; on a separate cybersecurity evaluation, Large 4 outscored two rivals. Its weights were still pending.