OpenAI admits Pro plan rate-limit bug affecting ~40,000 accounts, offers compensation

OpenAI has acknowledged a usage-limit bug: users who upgraded from Pro 200 ($200/month) to Pro 500 ($500/month) did not get higher weekly Codex and Work quotas—limits kept running under the old plan, meaning "the $500 tier was effectively running as $200." The issue was fixed on October 16, and affected users are gradually receiving compensation.

Confirmed

Several Reddit users shared official OpenAI emails confirming that post-upgrade quotas were incorrectly calculated as if they were on Pro 500. According to one account, OpenAI engineering lead Thibault Sottiaux said roughly 40,000 accounts had a mismatch between their enforced limits and their chosen plan.
OpenAI fixed the issue on October 16 and said it has been contacting affected users one by one.
The compensation is 25,000 credits (worth about $1,000); the email cited in the m5 post matches the $1,000 figure reported by multiple users.
The m5 post author did the math, noting that Pro 500 users had actually been getting the same quota as Pro 200, so they paid far more than the service delivered.

Why it matters

The incident involves paying subscribers not getting what they paid for, directly harming the core interests of high-paying Pro users. The compensation (about $1,000)—more than double the monthly plan price—signals how seriously OpenAI is taking the mistake.
With around 40,000 accounts affected, this is not an isolated case but a systemic quota-enforcement bug, posing a challenge to trust in subscription services and reminding users to verify their actual quotas after upgrading.

2026-10-20 ~ 2026-10-21 · 6 related posts

Primary sources

OpenAI admits Pro 500 allowance bug, compensates user with $1,000 in credits — XFreeze · 2026-10-21 00:06 (UTC+08:00)
OpenAI admits Pro plan allowance bug, hands user $1,000 in credits — xXDrDreadXx · 2026-10-20 23:22 (UTC+08:00)
OpenAI admits Codex allowance drop was an error fixing Pro 200 downgrades, gives $500 credits — cneuralnetwork · 2026-10-21 01:40 (UTC+08:00)

[source] OpenAI admits Pro plan allowance bug, hands user $1,000 in credits — xXDrDreadXx · 2026-10-20
[source] OpenAI admits Pro 500 allowance bug, compensates user with $1,000 in credits — XFreeze · 2026-10-21
$500 OpenAI Pro plan ran with $200-tier limits; affected users get $1k in credits — Angaisb_ · 2026-10-21
OpenAI says usage-limit bug hit ~40,000 accounts, will compensate with credits — pvncher · 2026-10-21

2 near-duplicate retellings: airesearch12 · cneuralnetwork