Lindsay McCallum Rémy RT @paolino: I've been watching this since the first Codex dropped. There's no comparison. Codex is miles ahead of Claude Code. But the narrative won't stop. Which tells me people didn't actually try Codex before confidently declaring a winner. Even Opus 4.6 on max effort hallucinates constantly, confidently. Not on maxed out contexts, on fresh conversations. Today: hallucinated results it should have searched. Blindly changed one without verifying. Searched but got the wrong answer. Then doubled down on a mistake it already made. Just a mess. 20 minutes for a simple row of feature comparisons. Codex: searched everything, found nuance, suggested ways to represent it in the table. One shot. tweet