本页为机器翻译。如有错误，请发邮件至 [email protected]。
$ git log --oneline limits/
变更日志
额度一有变动就记在这里。RSS 订阅。第三方报告带有 第三方报告 标记。 条目保留英文原文。
额度变动时提醒我
每次真实变动发一封简短邮件，别无其他。
你的邮箱和选择由我们的邮件服务商 Resend 保存。绝不出售。每封邮件都可退订。
OpenAI
即将生效
Upcoming: grandfathered Pro 200 allowance ends (reported 20x -> 10x Plus)
Multipliers from third-party/X post, not official page
GitHub
Copilot deprecated GPT-5.5, GPT-5.4, GPT-5.4 mini, GPT-5 mini, Gemini 3.7 Flash and Grok 4.5 across most Copilot experiences (suggested replacements: GPT-6.1 Sol, GPT-6 Luna, Gemini 3.8 Flash, Grok 4.7)
Announced Sep 18, 2026. Some GitHub plan pages still list a few of these models (e.g. Grok 4.5, Gemini 3.7 Flash) as of Oct 20.
OpenAI
GPT-5.5 retired from Codex with ChatGPT sign-in (noon PDT); API unaffected
Cursor
第三方报告
$20 Pro closed to new signups; new Plus $8/Premium $30/Super $100/Ultra $200; 3rd-party models billed at API list price
Current plans confirmed on cursor.com/pricing; date third-party
Google
Gemini Code Assist Standard and Enterprise end of sale: new subscriptions can no longer be purchased; existing ones auto-renew through 2026, auto-renewal ends in 2027
Google points to Antigravity (via Gemini Enterprise / Gemini Enterprise Agent Platform) as the alternative.
Anthropic
OpenAI
Pro 200 reopened to new subs with lower allowance; grandfathered (active Sep 22-29) keep old allowance through 2026-10-29; Pro 500 launched
OpenAI
Automatic Codex usage reset for Plus/Pro/Business
Google
New Gemini Code Assist subscriptions no longer purchasable in the Cloud console for billing accounts without an active subscription (sales contact only)
OpenAI
Banked Codex resets granted with GPT-6 Astra rollout
Z.ai
GLM Coding Plan switched to credits (5-hour + weekly credit caps, off-peak at half rate); old prompt-based plans closed to new users
Existing V1/V2 and team subscribers keep their old limits and can renew. Weekends are now off-peak all day for existing plans too.
GitHub
Vercel
Google
Gemini CLI free tier retired: unpaid (Code Assist for individuals) and Google One users moved to Antigravity CLI; Code Assist IDE extensions replaced by Antigravity
Stated in site banners on codeassist.google and geminicli.com/docs/resources/quota-and-pricing/; not in the GCA release notes.
Cognition
Windsurf renamed Devin Desktop; plans and prices unchanged, including legacy Windsurf plans
windsurf.com now redirects to devin.ai. On Teams, Devin Desktop needs a full seat ($40/mo).
GitHub
GitHub
Cursor
Cognition
Cognition
Windsurf Max drops its daily quota; weekly limit stays (Pro and Teams keep daily + weekly)
Alibaba Cloud
Model Studio Coding Plan Lite closed to new purchases (renewals and upgrades ended 2026-04-13)
Cognition
Windsurf replaces credits with daily + weekly token quotas: Pro $20, Teams $40/seat, new Max $200; extra usage at API prices
Announced Mar 18. Existing Pro ($15) and Teams ($30/seat) keep their price indefinitely; unused add-on credits became extra-usage balance at $10 per 250 credits.
Replit
GitHub
Alibaba Cloud
Coding Plan Pro no longer sold once sold out; the AI Coding Plan campaign page now redirects to Token Plan
No end date published. Redirect observed 2026-10-20.
MiniMax
M Plan (Go $22 / Explore $55 / Build $132) replaced Token Plan for new buyers; text limits unchanged, H3 video added on Explore/Build
Launch date not stated; the M Plan launch offer ran through Oct 14, 2026. Existing Token Plan subscribers keep their plan while auto-renew is on.
Cursor
Cursor Token Rate ($0.25 per million tokens on Teams and Enterprise third-party requests) no longer applies to 15 smaller third-party models, including Claude Haiku 5.5, Gemini 3.8 Flash, GPT-5.6 Luna, GPT-5.4 Mini and GPT-5 Mini
No effective date stated. First seen on Cursor’s pricing docs on 2026-10-21. The docs now also say the rate applies to input, output and cached tokens. Individual plans (including legacy Pro) never paid this rate, so they are unaffected.