OpenAI's Codex team confirmed today that it is looking into user reports of degraded performance from GPT-5.5. Team member Thibault Sottiaux said the underlying systems are running healthily and that the investigation hasn't reached any firm conclusions yet.
https://twitter.com/sama/status/2055356452286640630
CEO Sam Altman then shared the statement, praising the team's responsiveness. He pointed out that in past similar investigations, the problem often wasn't that the model got worse — it's that users got used to the existing capabilities and started demanding more.