It seems the model is also pretty good at catching other people’s mistakes. According to a report from Axios, Opus 4.6 was ...