the @xai team seems pretty out-of-touch with how far behind grok is on deeply technical work this advantage was clearly gained by oai and anthropic by shipping coding models which then got more granular data on codebases and industry-specific methodology, variables that matter, or just general assumptions. reasoning between models on technical questions here this is dramatically different. hopefully xai knows about this. i'm a fan of grok for other things and would like it to be competitive
this is with the newest grok 4.20 heavy btw
212