Codex Plus users report allowances draining ~20x faster with no settings change

coffeelover1010 · ghdev · 2026-10-10

A ChatGPT Plus user filed issue #52851 on openai/codex claiming their Codex desktop allowance now runs out in roughly one hour—about 20x faster than before—with no changes to model, plan, or settings, and three recent usage resets exhausted. The author admits this is anecdotal, not token-normalized, and offers no controlled reproduction, but points to similar reports (#41220 and historical #28879) and asks OpenAI to investigate usage accounting, entitlements, or unintended background activity. A billing-related complaint worth watching for paying users.

Original post →

More from coding & agent

coding & agent channel →