OpenAI’s Codex coding agent has seen usage-limit resets cluster around major model launches and high-demand periods, with the week of September 21 shaped by the September 22 rollout of GPT-6 Sol and Luna. The accompanying announcement from Codex lead Tibo included a banked reset credited to Plus, Pro, and Business accounts, following a pattern of discretionary resets issued for compensation, bug fixes, or traffic spikes. Traders see roughly even odds for zero to three such events because the exact count hinges on any follow-on issues with the new models, additional compensation decisions, or further announcements, while four or more remains unlikely given historical cadence. These market-implied odds reflect aggregated trader assessment of OpenAI’s response patterns rather than a fixed schedule.
Экспериментальная сводка, созданная ИИ на основе данных Polymarket. Это не является торговой рекомендацией и не влияет на то, как разрешается этот рынок. · ОбновленоСколько Codex сбрасывает за неделю до 21 сентября?
2 27%
3 26%
0 14%
1 14%
0
14%
1
14%
2
27%
3
26%
4+
2%
2 27%
3 26%
0 14%
1 14%
0
14%
1
14%
2
27%
3
26%
4+
2%
Resets will be counted by the calendar date (ET) on which OpenAI announces or initiates them, and at most one reset will be counted per calendar date. Announcements, confirmations, or phased rollouts of a reset already counted will not add to the total.
A qualifying reset is a discretionary, non-scheduled action by OpenAI that, ahead of the normal reset schedule, either zeroes or refills the weekly or generally applicable usage allowance for paid Codex or ChatGPT Work subscription users, or grants those users a reset that OpenAI characterizes as a banked or redeemable rate-limit reset, regardless of whether or when individual users redeem it. A reset will still qualify if certain models with separate usage limits, certain plans, or some users are not covered due to technical exceptions, phased rollout, or plan differences.
Normal scheduled resets, resets granted only to individually selected users or accounts, temporary limit increases, additional usage or credits that do not refill existing allowances, and changes to pricing, plans, or usage caps will not qualify. Resets of OpenAI API rate limits or organizational usage tiers will also not qualify. Only qualifying resets announced or initiated within the specified week will be counted.
An official announcement stating that OpenAI has reset, is resetting, or will reset the limits on the same date as the announcement will count if published within the specified week, regardless of when the technical reset begins or is completed. Qualifying announcements include X posts by official OpenAI accounts or OpenAI employees authorized to announce Codex usage changes, for example: https://x.com/OpenAI/status/2042296046009626989 and https://x.com/sama/status/2041658719839383945.
This market will resolve to "No" as soon as the specified number of resets becomes impossible, for example because more resets than the specified number have already occurred.
The primary resolution source for this market will be official communications from OpenAI, including X posts published by @OpenAI, @OpenAIDevs, @sama, @thsottiaux, or any other OpenAI-operated account, as well as official OpenAI blog posts, changelogs, and in-product notices. However, a consensus of credible reporting may also be used.
Открытие рынка: Sep 22, 2026, 11:04 AM ET
Кто определяет исход
0x69c47De9D...Resets will be counted by the calendar date (ET) on which OpenAI announces or initiates them, and at most one reset will be counted per calendar date. Announcements, confirmations, or phased rollouts of a reset already counted will not add to the total.
A qualifying reset is a discretionary, non-scheduled action by OpenAI that, ahead of the normal reset schedule, either zeroes or refills the weekly or generally applicable usage allowance for paid Codex or ChatGPT Work subscription users, or grants those users a reset that OpenAI characterizes as a banked or redeemable rate-limit reset, regardless of whether or when individual users redeem it. A reset will still qualify if certain models with separate usage limits, certain plans, or some users are not covered due to technical exceptions, phased rollout, or plan differences.
Normal scheduled resets, resets granted only to individually selected users or accounts, temporary limit increases, additional usage or credits that do not refill existing allowances, and changes to pricing, plans, or usage caps will not qualify. Resets of OpenAI API rate limits or organizational usage tiers will also not qualify. Only qualifying resets announced or initiated within the specified week will be counted.
An official announcement stating that OpenAI has reset, is resetting, or will reset the limits on the same date as the announcement will count if published within the specified week, regardless of when the technical reset begins or is completed. Qualifying announcements include X posts by official OpenAI accounts or OpenAI employees authorized to announce Codex usage changes, for example: https://x.com/OpenAI/status/2042296046009626989 and https://x.com/sama/status/2041658719839383945.
This market will resolve to "No" as soon as the specified number of resets becomes impossible, for example because more resets than the specified number have already occurred.
The primary resolution source for this market will be official communications from OpenAI, including X posts published by @OpenAI, @OpenAIDevs, @sama, @thsottiaux, or any other OpenAI-operated account, as well as official OpenAI blog posts, changelogs, and in-product notices. However, a consensus of credible reporting may also be used.
Кто определяет исход
0x69c47De9D...OpenAI’s Codex coding agent has seen usage-limit resets cluster around major model launches and high-demand periods, with the week of September 21 shaped by the September 22 rollout of GPT-6 Sol and Luna. The accompanying announcement from Codex lead Tibo included a banked reset credited to Plus, Pro, and Business accounts, following a pattern of discretionary resets issued for compensation, bug fixes, or traffic spikes. Traders see roughly even odds for zero to three such events because the exact count hinges on any follow-on issues with the new models, additional compensation decisions, or further announcements, while four or more remains unlikely given historical cadence. These market-implied odds reflect aggregated trader assessment of OpenAI’s response patterns rather than a fixed schedule.
Экспериментальная сводка, созданная ИИ на основе данных Polymarket. Это не является торговой рекомендацией и не влияет на то, как разрешается этот рынок. · Обновлено

Не доверяй внешним ссылкам.
Не доверяй внешним ссылкам.
Часто задаваемые вопросы