Reactions from ranked influencers
3 posts
gpt 5.6 pro is substantially smarter than 5.6 in codex. it is inordinately good at critiquing and fixing 5.6's dumb ideas and implementations. there must be a better way than asking 5.6 codex to write research reports, me pasting into pro, then copying instructions back!
i think we need 5.6 pro in codex, and we need some integrated system to critique and review what the agent is off doing every 30 minutes. otherwise agents get lost and waste time and do weird stuff without this
@willdepue EZ have it use the browser on its own