THE SHORT VERSION
- OpenAI said it is doubling usage limits for the ChatGPT Pro $100, $200 and $500 tiers and Business Premium, effective immediately.OpenAI 宣布把 ChatGPT Pro 100、200、500 美元三档及 Business Premium 的用量上限翻倍，即时生效。
- The company had planned to reduce included usage for Pro 200 subscribers after October 29 while keeping the price unchanged, and reversed that after feedback.公司原计划在 10 月 29 日后下调 Pro 200 的包含用量、但维持 200 美元月费，在用户反馈后撤销。
- The change affects ChatGPT Work and Codex, which share usage allowances; OpenAI said it will keep extended limits for its background assistants, called Dots.该变化主要影响 ChatGPT Work 与 Codex（两者共享额度）；OpenAI 称后台助手 Dots 暂继续享有扩展上限，且用量独立计算。
What OpenAI announcedOpenAI 宣布了什么
OpenAI has doubled the usage limits on its paid ChatGPT tiers instead of cutting them. Android Authority reported that the change covers the ChatGPT Pro plans priced at $100, $200 and $500 a month as well as Business Premium, and that it takes effect immediately.OpenAI 把付费 ChatGPT 各档的用量上限翻倍，而不是下调。Android Authority 报道称，此次调整覆盖每月 100、200、500 美元的 ChatGPT Pro 各档以及 Business Premium，并即时生效。
BleepingComputer reported the same announcement and the detail that matters for most subscribers: the higher allowances primarily affect ChatGPT Work and Codex, which share usage limits under eligible subscriptions.BleepingComputer 报道了同一公告，并给出对多数订阅者最要紧的细节：更高的额度主要影响 ChatGPT Work 与 Codex，两者在符合条件的订阅中共享用量上限。
The cut that did not happen那次没有发生的降额
The announcement is a reversal, not an upgrade. Both outlets reported that OpenAI had planned to reduce the included usage allowance for existing Pro 200 subscribers after October 29, without lowering the $200 monthly price. That combination, less service for the same money, is the kind of change subscribers notice.这次公告是一次撤销，而不是一次升级。两家媒体都报道称，OpenAI 原计划在 10 月 29 日之后下调现有 Pro 200 订阅者的包含用量，却不降低每月 200 美元的价格。花同样的钱得到更少的服务，正是订阅者会注意到的变化。
The company then said it had heard the feedback and would not reduce the Pro 200 limits. In a post quoted by both outlets, it said usage limits for all Pro 100, 200 and 500 tiers and Business Premium would be doubled, effective today, and that the Work and Codex usage for Pro 200 would stay the same rather than shrink.随后公司表示已听到反馈，不会削减 Pro 200 的上限。两家媒体引用的帖子称，100、200、500 美元三档与 Business Premium 的用量上限都将翻倍、即时生效，且 Pro 200 的 Work 与 Codex 用量将保持不变而不会缩小。
Why Codex and Work are the pressure point为什么压力点在 Codex 与 Work
The tiers are not generic chatbot plans any more. Codex is a coding agent, and Work is the version of ChatGPT packaged for organisations; both consume far more tokens per task than a conversation. A user who runs an agent that reads a repository, edits files and iterates will hit an allowance that a chat user never approaches.这些档位已不再是通用聊天机器人套餐。Codex 是编程智能体，Work 是面向组织打包的 ChatGPT 版本；两者每个任务消耗的 token 都远高于一次对话。用户若运行一个读取代码库、修改文件并反复迭代的智能体，就会撞上聊天用户永远碰不到的额度。
That is why a limit change is felt most by the customers paying the most. It is also why doubling the allowance is a cheaper concession than it looks: the heavy users are a minority, and the cost of letting them run more is concentrated in exactly the compute the company is already buying in bulk.这也是为什么额度变化对支付最多的客户感受最强。同样，翻倍是一项比看起来更便宜的让步：重度用户是少数，让他们多跑的代价集中在公司本就批量采购的算力上。
The Dots exceptionDots 的例外
Alongside the subscription change, OpenAI said it would continue offering extended limits for Dots, its always-on background assistants, at least for now, and that usage of a Dot remains separate from a subscriber's default usage. That separation matters: a background assistant that keeps running would otherwise drain the same allowance a person uses interactively.在订阅调整之外，OpenAI 称其常驻后台助手 Dots 暂时继续享有扩展上限，且 Dot 的用量仍与订阅者的默认用量分开计算。这种分离很重要：持续运行的后台助手否则会消耗掉用户交互所用的同一份额度。
The word "for now" is doing work in that sentence. It tells subscribers the exception is temporary without saying what replaces it, and it leaves the company room to fold Dots into the standard allowance later. Anyone building a workflow around a Dot should read that as a warning rather than a guarantee.「暂时」一词在这句话里很关键。它告诉订阅者这一例外是临时的，却没有说明替代方案，也给自己留出了日后把 Dots 并入标准额度的空间。围绕 Dot 搭建工作流的人，应把这当作提醒而不是保证。
The numbers are still not published具体数字仍未公布
BleepingComputer noted that OpenAI will not provide exact numerical usage limits for each tier. That is consistent with how the company has handled allowances: the plans are described in multiples and qualitative terms, and the actual ceiling is discovered by hitting it.BleepingComputer 指出，OpenAI 不会提供各档确切的数值上限。这与该公司一贯处理额度方式一致：套餐以倍数与定性描述呈现，真实天花板要靠撞上去才知道。
For a subscriber, that makes announcements like this hard to compare across providers. A doubled limit is good news, but a buyer cannot calculate whether it is enough without knowing the starting point, and the company has chosen not to give one. The practical advice is to watch the usage dashboard rather than the press release.对订阅者而言，这让此类公告很难跨厂商比较。上限翻倍是好消息，但买家在不知道起点的情况下无法计算它是否够用，而公司选择不给起点。实用建议是看用量面板，而不是看新闻稿。
What a limit change signals额度变化说明了什么
Usage limits are how a subscription AI business manages capacity without publishing prices per token. When compute is tight, the cheapest lever is to reduce what a flat fee includes; when competition is tight, the cheapest lever is to raise it. A doubling followed by an apology for a cut is a company saying both constraints bind at once.用量上限，是订阅制 AI 业务在不公布每 token 价格的情况下管理容量的一种方式。算力紧张时，最省力的杠杆是削减固定费用所包含的内容；竞争紧张时，最省力的杠杆是提高它。先翻倍、再为一次降额道歉，等于公司同时承认两种约束都在起作用。
It is also a reminder that the customer is buying capacity, not software. A feature can be shipped once and used by everyone; a token allowance is consumed every time somebody runs an agent. That makes the allowance the part of the product most sensitive to the cost of compute, and the part most likely to change again.这也提醒人们：客户买的是容量，不是软件。功能可以发布一次、所有人共用；token 额度则每跑一次智能体就被消耗一次。这使额度成为产品中对算力成本最敏感的部分，也最可能再次变化。
How subscribers should read it订阅者该怎么读这条消息
The practical question for a buyer is not whether the limit doubled but whether the doubled limit covers the work. A team that runs Codex agents all day should measure its actual consumption against the allowance rather than against a competitor that publishes different units, because the units are not comparable across providers.对买家而言，实际问题不是上限有没有翻倍，而是翻倍后的上限是否覆盖工作。整天跑 Codex 智能体的团队，应拿自己的实际消耗对照额度，而不是对照另一家使用不同计量单位的厂商——不同厂商的单位无法直接比较。
Until OpenAI publishes numbers, the only reliable measurement is the one in the account. That is an unsatisfying way to buy a subscription, and it is the state of the market: plans are sold on qualitative promises while the real constraint is enforced silently and changed by announcement.在 OpenAI 公布数字之前，唯一可靠的度量是账户里的那一个。这不是购买订阅的令人满意的方式，却正是市场现状：套餐以定性承诺出售，而真正的约束被静默执行、并通过公告改变。
What is established, and what is not哪些已经确定，哪些还没有
Established: OpenAI doubled usage limits for the ChatGPT Pro $100, $200 and $500 tiers and Business Premium; the change took effect immediately; it primarily affects Work and Codex; the company reversed a planned cut for Pro 200 after feedback; extended Dots limits continue for now and remain separate. Not established: the exact numerical limits per tier, how long the Dots exception lasts, and whether the doubled allowances survive the next pricing review.已经确定的：OpenAI 把 ChatGPT Pro 100、200、500 美元三档与 Business Premium 的用量上限翻倍；即时生效；主要影响 Work 与 Codex；公司在反馈后撤销了 Pro 200 的降额计划；Dots 的扩展上限暂时延续且独立计算。尚未确定的：各档确切数值上限、Dots 例外持续多久，以及翻倍后的额度能否挺过下一轮定价评估。
For readers, the useful signal is that subscription pricing in this market is being set by usage rather than by features. This site has covered Anthropic's rename of its own Claude tiers and the enterprise agent partnerships that push the same allowances, and the scale of AI spending commitments explains why the compute behind them is the constraint everybody is managing.对读者而言，有用的信号是：这个市场的订阅定价正由用量而非功能决定。本站报道过Anthropic 对自家 Claude 档位的改名与推动同类额度的企业智能体合作，而AI 支出承诺的规模解释了为什么背后的算力是所有人都在管理的约束。
Sources & reporting limits来源与核查范围
Follow Industry to find related stories on your next visit.关注产业动态，下次访问时阅读相关文章。
