[WhiteWorld]

Fixes #149

Review date for verification is 2026-11-03 (Asia/Shanghai). Both files now have last_full_review: 2026-11-03 and next_review_due: 2026-12-03. npm run content:check -- --strict reports no errors. Scores are unchanged.

Ark Coding Plan

Configurable Model Name list on the get-started page now includes kimi-k3.1-pro (feature release 10.24) and no longer includes minimax-m3 or kimi-k2.7-code. Both shutdown times are in the past (minimax-m3 2026-11-02 14:00:00 UTC+8, kimi-k2.7-code 2026-10-28 14:00:00 UTC+8), so supported_models follows that live list and drops both IDs.

Discount rules page docs/82379/2525065 (page UpdatedTime 2026-07-27, still published on 2026-11-03) lists Lite 40 / 9.9 yuan and Pro 200 / 49.9 yuan. The window is 2026-06-08 through 2026-11-08, at most two months per account, then list price. A 9.5折 referral discount can stack. The page does not print a combined yuan amount, so the old ¥8.9 figure is no longer stated as current. The three-month half-price claim was not on this page.

The roughly 1,200 / 6,000 requests per 5 hours were not on any page opened this pass. They stay in pricing.note as the earlier record, and they are no longer shown on the coding-plan card. The activity page does say Pro usage is about 5× Lite.

Pages opened

• https://ark.volcengine.com/region:cn-beijing/docs/coding-plan-personal-feature-release
• https://ark.volcengine.com/region:cn-beijing/docs/coding-plan-personal-model-deprecation
• https://ark.volcengine.com/region:cn-beijing/docs/coding-plan-personal-plan-overview
• https://ark.volcengine.com/region:cn-beijing/docs/coding-plan-personal-get-started
• https://ark.volcengine.com/region:cn-beijing/docs/coding-plan-personal-model-release
• https://ark.volcengine.com/region:cn-beijing/docs/coding-plan-personal-model-discount
• https://www.volcengine.com/docs/82379/2525065
• https://www.volcengine.com/docs/82379/1925114 (套餐概览 mirror, UpdatedTime 2026-10-29; no request counts)
• https://www.volcengine.com/L/s3lNTNYxaEc/

Console was not opened. current_version stays the old label v2. Confidence stays medium.

Zhipu GLM Coding Plan

Subscribe page still shows Lite/Pro/Max monthly prices ¥118 / ¥538 / ¥1,078, and also shows ¥94.4 / ¥430.4 / ¥862.4, each labeled per month. The usage page still lists 2,000/10,000, 12,000/60,000, and 28,000/140,000 credits. Models remain GLM-5.3 and GLM-5.3-Flash. OX Alpha is the internal codename for Flash, not a new model.

The tools page now lists 23 allowed clients, including ZCode and AutoClaw at 1.5× quota. OpenClaw and other general agents are secondary scheduling. Confidence stays high for the monthly prices and credit quotas.

Pages opened

• https://www.bigmodel.cn/glm-coding
• https://docs.bigmodel.cn/cn/coding-plan/overview
• https://docs.bigmodel.cn/cn/coding-plan/notice/usage-revision
• https://docs.bigmodel.cn/cn/coding-plan/tool/others
• https://docs.bigmodel.cn/cn/coding-plan/faq
• https://docs.bigmodel.cn/cn/coding-plan/quick-start

For Bai

1. Ark activity-page Pro monthly card has priceConfig 9.9 and 40/月 while the SKU is Coding_Plan_Pro_monthly. Quarterly strikethrough is 600/季 and annual is 2400/年, which match a ¥200 list price. This file keeps Pro at ¥200 from the discount-rules table and does not adopt the duplicated 9.9 card.
2. Quarterly and annual tab image filenames say 4.9折 and 8.7折. The card amounts (Lite 116.4/季 and 456/年, Pro 582/季 and 2280/年) do not match those rates. They are recorded as raw card numbers only, not as current discount rates. Annual 456 and 2280 do equal 95% of the list totals.
3. Deprecation docs have not caught up. The retired table does not list minimax-m3 or kimi-k2.7-code yet. The plan overview still marks them 即将下线. minimax-latest and kimi-latest still point at them. deepseek-latest is deepseek-v4.1-flash on the deprecation page and deepseek-v4-pro on the model-release page.
4. The deduction page dates kimi-k3 through 2026-10-25 but its status table still says 进行中. The plan overview's current promo bullets do not include kimi-k3. This pass does not treat kimi-k3 as still discounted.
5. Ark request counts (about 1,200 / 6,000 per 5 hours) are still unverified. Timezone is still unnamed on the 5-hour / weekly / monthly reset rules.
6. Ark tool cards say TraeCode. This file still says Trae. OpenViking is in the March feature note and not on the current tool cards. Client list was left as-is.
7. Zhipu annual 7折 amounts (¥82.6 / ¥376.6 / ¥754.6) were not in the pages opened this pass. They remain in the plan notes only. best_for no longer recommends the annual discount as a current reason to buy. The subscribe page's ¥94.4 / ¥430.4 / ¥862.4 match the old quarterly 8折 equivalents, but the extracted text never said 包季 or 8折.
8. The Zhipu marketing page also names TraeWork and TraeCode. The tools restriction page only lists TRAE, so those two names were not added.
9. Guides and compare articles still describe the two Ark shutdowns as upcoming and still say the ¥9.9 / ¥49.9 promo may belong to Agent Plan. This PR does not edit those articles.

Checks

• npm run content:check -- --strict passes (0 errors).
• npm run build completes.
• Built /zh/coding-plan/ shows Ark Lite ¥40 / Pro ¥200, model tags starting with kimi-k3.1-pro, and no minimax-m3 or kimi-k2.7-code on that card. Zhipu shows ¥118 / ¥538 / ¥1,078 with the credit quotas and the expanded client list. Desktop and a 390px viewport keep the price text inside the cards.

  

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
codepick	Ready	Preview	Nov 3, 2026 2:25am UTC

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-03T02:50:26.701749Z	58b93ef	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.