OpenAI adds Luna Reserve fallback for Codex usage limits

kimmonismus · x · 2026-08-27

OpenAI appears to have introduced 'Luna Reserve' for Codex, a capped fallback mechanism that activates when advanced model limits are reached. This aims to balance user-friendliness with abuse prevention, replacing the previous exploitable 'run past 0%' behavior and the disruptive hard stops.

Original post →

More from Models

Models channel →