“@wholemars I must give @AnthropicAI credit here: Opus 4.5 is outstanding. They are doing particularly excellent work at the pretraining level. However, Grok wins when applied to logic (software is just logic at scale) that isn’t in the Opus pretraining. For example, the Tesla chip design… https://t.co/JkpPOYGJIN https://x.com/i/web/status/1998876157991952622”