[{"data":1,"prerenderedAt":57},["Reactive",2],{"news-story-news-20261025-kimi-k3-1-pro-plan-quota-reset-apology":3},{"id":4,"title_zh":5,"title_en":6,"summary_zh":7,"summary_en":8,"takeaways_zh":9,"takeaways_en":15,"content_zh":21,"content_en":22,"external_links":23,"source":32,"category":39,"impact_score":40,"views_count":41,"likes_count":42,"related_entities":43,"tags":50,"image_url":56},"news-20261025-kimi-k3-1-pro-plan-quota-reset-apology","Kimi 官方就 K3.1 Pro 订阅限流与额度误耗致歉：全体订阅用户一次性返还 25% 月度额度，并下调子 Agent 调用与输出长度","Kimi apologizes for K3.1 Pro plan rate limits and wasted quota, gives every subscriber a one-time 25% monthly quota reset and trims subagent calls and output length","月之暗面官方账号 @Kimi_Moonshot 在 UTC 2026-10-25 20:14 发帖（[原帖](https://x.com/Kimi_Moonshot/status/2114450610724708492)），承认 K3.1 Pro 订阅计划出现限流和不必要的额度消耗，称「这是我们的问题，本应更早修复」。补偿方案：所有计划的每位订阅用户获得一次性用量重置，返还 25% 月度额度（已用不足 25% 的直接清零），约等于一周用量，用量条将在太平洋时间当天结束前更新。官方同时调整了计划参数，减少不必要的子 Agent 调用和输出长度，让额度更耐用。","At 20:14 UTC on 2026-10-25, Moonshot AI's official account @Kimi_Moonshot ([post](https://x.com/Kimi_Moonshot/status/2114450610724708492)) acknowledged rate limits and unnecessary quota consumption on the K3.1 Pro subscription plan, saying \"these were our issues, and we should have fixed them sooner.\" Every subscriber on every plan gets a one-time usage reset that restores 25% of their monthly quota (usage below 25% resets to zero), equal to about one week of usage; usage bars update by end of day PST. Kimi also adjusted the plan's parameters and is cutting unnecessary subagent calls and output length so quota lasts longer.",[10,11,12,13,14],"补偿力度：一次性返还 **25%** 月度额度，约等于 **1 周**用量（[官方原帖](https://x.com/Kimi_Moonshot/status/2114450610724708492)）","已用不足 25% 的用户用量直接清零，覆盖**所有**订阅计划","生效时间：用量条在太平洋时间 10-25 当天结束前更新（约新加坡时间 10-26 15:00）","根因方向：官方称将减少不必要的子 Agent 调用和输出长度，并调整计划参数","背景：K3.1 Pro 于 10-23 发布，API 价格 $2/$10 每百万 token（官方口径）",[16,17,18,19,20],"Compensation: a one-time **25%** monthly quota restore, about **1 week** of usage ([official post](https://x.com/Kimi_Moonshot/status/2114450610724708492))","Users below 25% usage are reset to zero; applies to **every** subscription plan","Timing: usage bars update by end of day PST on Oct 25 (about 15:00 SGT on Oct 26)","Fix direction: fewer unnecessary subagent calls, shorter outputs and adjusted plan parameters","Context: K3.1 Pro launched Oct 23 with API pricing of $2/$10 per million tokens (vendor figures)","### 核心背景与行业痛点\nKimi K3.1 Pro 在 10 月 23 日发布后，订阅用户（主要通过 Kimi Code 等编程工具使用）很快遇到两个问题：频繁被限流，以及额度被不必要地快速消耗。编程 Agent 的订阅制普遍按额度计费，一次长任务会触发大量子 Agent 调用和长输出，计费逻辑稍有偏差，用户的月度额度就会提前见底。厂商如何回应，直接影响开发者是否续订。\n\n### 架构亮点与底层机制\n官方原帖给出的处理分两部分。第一是补偿：所有计划的每位订阅用户获得一次性用量重置，返还 25% 月度额度，已用不足 25% 的直接清零，官方称约等于一周用量。第二是修正消耗来源：调整计划参数，并减少不必要的子 Agent 调用和输出长度。官方没有公布具体改了哪些参数，也没有给出限流阈值的具体数字。\n\n### 权威 Benchmark 与实测跑分对比\n这次是计费与服务调整，不涉及模型能力变化，官方未公布任何跑分或额度消耗前后的对比数据。可作参照的是发布时的官方口径：K3.1 Pro API 价格为输入 $2、输出 $10 每百万 token。减少子 Agent 调用和输出长度后，同样任务的实际额度消耗会降多少，目前只能等用户实测。\n\n### 开发者实战落地与开箱指南\n订阅用户不需要做任何操作，官方称用量条会在太平洋时间 10 月 25 日当天结束前更新，大约是新加坡时间 10 月 26 日下午。建议更新后先核对剩余额度是否按 25% 返还；跑长程任务时留意子 Agent 数量和输出长度是否明显减少；如果仍频繁限流，可对比按量计费的 API 是否更划算。","### Background and pain point\nAfter Kimi K3.1 Pro launched on October 23, subscribers, many using it through coding tools such as Kimi Code, ran into frequent rate limits and quota that drained faster than it should. Coding-agent subscriptions are usually metered by quota, and a long task can fire many subagent calls and long outputs, so a small metering problem can exhaust a monthly allowance early.\n\n### What changed\nThe official post has two parts. First, compensation: every subscriber on every plan gets a one-time usage reset that restores 25% of the monthly quota, and anyone below 25% usage is reset to zero, which Kimi says equals about one week of usage. Second, a fix to the source of the drain: plan parameters were adjusted and unnecessary subagent calls and output length are being reduced. Kimi did not say which parameters changed or publish rate-limit thresholds.\n\n### Benchmarks\nThis is a billing and service change, not a model change, and Kimi published no benchmarks or before-and-after consumption data. For reference, the launch-time vendor pricing for the K3.1 Pro API is $2 input and $10 output per million tokens. How much the trimmed subagent calls and outputs save on real tasks is still unmeasured.\n\n### What developers should do\nSubscribers do not need to act; Kimi says usage bars update by end of day PST on October 25, roughly the afternoon of October 26 in Singapore. Check that 25% was restored, watch whether long tasks spawn fewer subagents and shorter outputs, and if rate limits persist, compare against pay-as-you-go API pricing. Source: [Kimi official post](https://x.com/Kimi_Moonshot/status/2114450610724708492).",[24,28],{"title":25,"url":26,"type":27},"Kimi Code (official product page)","https://www.kimi.com/code","other",{"title":29,"url":30,"type":31},"Kimi K3.1 Pro tech blog (official)","https://www.kimi.ai/blog/kimi-k3-1-pro","blog",{"platform":33,"author_name":34,"author_handle":35,"author_avatar":36,"tweet_url":37,"published_at":38},"x","Kimi.ai","@Kimi_Moonshot","https://unavatar.io/x/Kimi_Moonshot","https://x.com/Kimi_Moonshot/status/2114450610724708492","2026-10-25T20:14:44.000Z","pricing",82,165,0,{"models":44,"agents":46,"vendors":48},[45],"Kimi K3.1 Pro",[47],"Kimi Code",[49],"Moonshot AI",[51,52,47,53,54,55],"Kimi","K3.1 Pro","quota","rate limits","subscription","",1793718870868]