Change log
44 of 44 dated changes. Newest first.
GPT-5.5 retired from Codex for ChatGPT sign-ins
GPT-5.5 selectable in Codex with ChatGPT sign-in.GPT-5.5 removed; GPT-6.1 Sol recommended on paid plans.Details and source
From 19:00 UTC on Oct 14, GPT-5.5 is no longer available in Codex when signed in with ChatGPT. Paid plans should switch to GPT-6.1 Sol; API-key Codex sessions are unaffected. Update saved settings, configs and scripts that select gpt-5.5.
Vendor’s framing: “The OpenAI API and Codex sessions authenticated with an API key aren’t affected.”
Source: ChatGPT & Codex changelog · Announced: 2026-09-14 (notice) and 2026-10-14, ChatGPT & Codex changelog · Plans: ChatGPT Plus (Codex), ChatGPT Pro 100, ChatGPT Pro 200Claude Code adds a ‘Wrap-Up Allowance’ at the five-hour limit
Responses stopped when the five-hour limit was reached.Small, capped wrap-up allowance (Claude Code 2.1.277+, gradual rollout).Details and source
If a response is in progress when you hit the five-hour limit, Claude Code may keep working briefly to reach a stopping point. Pro: up to once per weekly period; Max/Team Premium: each time. Counts toward the weekly limit.
Vendor’s framing: “a small, capped amount of additional usage that varies by plan. It is not unlimited”
Source: Claude Help Center · Announced: 2026-10-14, @ClaudeDevs; terms in Claude Help Center · Plans: Claude Pro, Claude Max 5x, Claude Max 20xCursor joins The X Subscription: one weekly pool shared with Grok
Monthly ‘Cursor Models’ + ‘Other Models’ pools (Pro $20, Pro+ $60, Ultra $200).New plans: weekly shared pool; Premium 4x, Super 15x, Ultra 50x Plus. Legacy plans unchanged unless you switch.Details and source
New customers get Free / Plus $8 / Premium $30 / Super $100 / Ultra $200 with one weekly pool for Cursor models shared with Grok; third-party models are no longer included (billed at API list price on Premium+). Legacy plans keep two monthly pools.
Vendor’s framing: “Each X Subscription plan includes at least twice as much coding usage with our models as the comparable Cursor plan.”
Source: Cursor docs · Announced: 2026-10-12, xAI news post and Cursor docs · Plans: Cursor on The X Subscription (Plus / Premium / Super / Ultra), Cursor Pro (legacy)Claude Max adds monthly API credits ($100 / $200), not usable in Claude Code
No API credits included with Max.$100 (Max 5x) / $200 (Max 20x) monthly API credits; plan usage limits unchanged.Details and source
Max 5x gets $100 and Max 20x gets $200 a month in Claude Platform API credits (Team: $20 per Standard seat, $100 per Premium seat, pooled up to $500). They cover the API, Managed Agents and the Agent SDK, but not interactive Claude Code or extra usage. No rollover; claim after 7 days on the plan.
Vendor’s framing: “The credits don’t change your usage limits in Claude, Claude Code, or Claude Cowork.”
Source: Claude Help Center · Announced: 2026-10-07, @ClaudeDevs; terms in Claude Help Center · Plans: Claude Max 5x, Claude Max 20xChatGPT Pro 200: lower included allowance for new subscribers
Previous Pro 200 included allowance.Lower included allowance (no figure published); grandfathered users move over after 2026-10-29.Details and source
New Pro 200 subscriptions get a lower usage allowance ‘to reflect our increasingly efficient models’. Existing subscribers (active Sep 22–29) keep the old allowance through Oct 29. Price stays $200.
Vendor’s framing: “a lower usage allowance than previously offered with Pro 200 to reflect our increasingly efficient models”
Source: OpenAI Help Center · Announced: OpenAI Help Center, ‘About ChatGPT Pro tiers’ · Plans: ChatGPT Pro 200
Claude Code weekly limits settle at +25% (a 17% cut vs. the boost)
Weekly limit at index 150 (temporary +50%).Weekly limit permanently at index 125 (+25% over pre-promo baseline).Details and source
Weekly limits became permanently 25% above the pre-promo baseline. Anthropic: “Compared to today, this works out to a 17% reduction in weekly limits on Claude Code.”
Vendor’s framing: “we’re permanently raising standard weekly limits in Claude Code by 25%”
Source: BleepingComputer · Announced: 2026-08-29, @ClaudeDevs on X · Plans: Claude Pro, Claude Max 5x, Claude Max 20xCopilot Business/Enterprise promo credits end
3,000 (Business) / 7,000 (Enterprise) AI credits per user per month.1,900 (Business) / 3,900 (Enterprise) AI credits per user per month.Details and source
The first-three-months promotional allowance (3,000 Business / 7,000 Enterprise credits per user) ended; standard is 1,900 / 3,900 per user per month, pooled.
Vendor’s framing: Scheduled end of a stated promotional window
Source: GitHub Docs · Announced: GitHub billing documentation (promo window 2026-06-01 to 2026-09-01) · Plans: GitHub Copilot Business / EnterpriseClaude Code +50% weekly boost extended to Sep 13
+50% weekly, ending 2026-08-31.+50% weekly, ending 2026-09-13.Details and source
The 50% boost stayed in place until the new permanent level took over on Sep 14.
Vendor’s framing: “Until then, the current 50% increase will be in place.”
Source: BleepingComputer · Announced: 2026-08-29, @ClaudeDevs on X · Plans: Claude Pro, Claude Max 5x, Claude Max 20xChatGPT Plus: five-hour Codex limit is back
No five-hour limit on Plus.Rolling five-hour limit on Plus.Details and source
The rolling five-hour limit returned for Plus across ChatGPT Work and Codex; Pro $100 and $200 exempt.
Vendor’s framing: “Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex. I had mentioned this a while ago, but then postponed it”
Source: Digital Applied ledger (row 19; original report: explainx) · Announced: 2026-08-25, Tibo Sottiaux on X · Plans: ChatGPT Plus (Codex)Cursor raises Cursor-models limits again (by email, no number)
Included usage as of the Jul 16 doubling.Higher included usage (magnitude not published).Details and source
Included usage limits for Cursor Models, including Auto, went up; announced only by email, no figure published.
Vendor’s framing: “Increased usage limits. Included usage limits for Cursor Models, including Auto, go up”
Source: Digital Applied ledger (row 21; original report: explainx) · Announced: On or about 2026-08-20, email to subscribers · Plans: Cursor Pro (legacy)Cursor Auto now billed at the routed model’s rate
Auto billed at one flat rate.Auto billed per routed model.Details and source
Auto stopped billing at one flat rate; each request bills at the rate of the model it is routed to, so the same pool can empty faster.
Vendor’s framing: Described as updated Auto pricing
Source: Digital Applied ledger (row 22; original report: Kilo) · Announced: On or about 2026-08-20, same email · Plans: Cursor Pro (legacy)Claude Code +50% weekly boost extended to Aug 31
+50% weekly, ending 2026-08-19.+50% weekly, ending 2026-08-31.Details and source
Fourth extension, with Anthropic saying it hoped to make the increase permanent but capacity may be tight.
Vendor’s framing: “We hope to make this a permanent change to our plans, but strong demand for our models means that capacity may be tight over the coming weeks.”
Source: ScriptXeno · Announced: 2026-08-19, @ClaudeDevs on X · Plans: Claude Pro, Claude Max 5x, Claude Max 20xZ.ai: 1.5x quota boost in ZCode through Aug 31
Standard credit quota.1.5x boost, 2026-08-14 to 2026-08-31.Details and source
Limited-time 1.5x quota boost for ZCode users, stated as up to 180% of standard quota with cache effects.
Vendor’s framing: A limited-time boost
Source: Digital Applied ledger (row 34; original: z.ai/blog/glm-5.3) · Announced: 2026-08-14, Z.ai GLM-5.3 launch post · Plans: Z.ai GLM Coding Plan (Lite / Pro / Max)GLM Coding Plan moves to credits
Prompts with five-hour and weekly caps.Credits per five hours and per week.Details and source
New subscribers get token-based credits: Lite 2,000/10,000, Pro 12,000/60,000, Max 28,000/140,000 (5-hour/weekly); off-peak usage at 50%; weekends all-day off-peak for existing subscribers.
Vendor’s framing: “To make usage clearer and easier to predict, GLM Coding Plan is moving to a credits-based system.”
Source: Z.ai docs · Announced: 2026-07-30, Z.ai Plan Update Announcement · Plans: Z.ai GLM Coding Plan (Lite / Pro / Max)Codex five-hour limit ‘restored’ (didn’t hold for Plus)
No five-hour limit.Five-hour limit restored as announced; whether it landed for Plus is not established.Details and source
OpenAI said the five-hour limit would return on Jul 30 with a full usage reset; on Aug 25 the same exec said he had postponed it.
Vendor’s framing: “Tomorrow, we’ll also restore the five-hour limit that we temporarily paused while investigating”
Source: Digital Applied ledger (row 18; original report: Codex Runway) · Announced: 2026-07-29, Tibo Sottiaux on X · Plans: ChatGPT Plus (Codex), ChatGPT Pro 100, ChatGPT Pro 200Kimi Code split announced — but not purchasable yet
One membership with monthly, weekly and five-hour limits.As announced: separate Kimi Code pool, monthly cap removed.Details and source
Moonshot announced a separate Kimi Code plan without a monthly total cap, but its own Help Center said the new plans were not yet available due to capacity.
Vendor’s framing: Framed as an increase
Source: Kimi Help Center (via Digital Applied ledger) · Announced: 2026-07-20, Kimi membership docsClaude Fable 5: Max keeps it (50% of weekly), Pro loses included access
Fable 5 included on all paid plans under a temporary access extension.Max/Team Premium: up to 50% of weekly limit. Pro/Team Standard: usage credits only.Details and source
Max and Team Premium get Fable 5 at up to 50% of the shared weekly limit; Pro and Team Standard lose included access (one-time $100 credit, then usage rates).
Vendor’s framing: Framed as Fable 5 becoming a permanent inclusion
Source: Digital Applied ledger (row 9) · Announced: 2026-07-18 to 2026-07-20, Anthropic X accounts and Help Center · Plans: Claude Pro, Claude Max 5x, Claude Max 20xClaude Code +50% weekly boost extended to Aug 19
+50% weekly, ending 2026-07-19.+50% weekly, ending 2026-08-19.Details and source
Second extension of the weekly promotion.
Vendor’s framing: “We’re also keeping Claude Code weekly limits 50% higher, now through August 19”
Source: ScriptXeno · Announced: 2026-07-18 to 2026-07-20 · Plans: Claude Pro, Claude Max 5x, Claude Max 20xQwen Coding Plan becomes a credits-based Token Plan
Pro ¥200/mo: 6,000 req/5h, 45,000/week, 90,000/month.Token Plan credits: Lite 700/2,500, Standard 3,000/10,000, Pro 12,000/40,000 (5h/7-day).Details and source
Request-count Coding Plan upgraded to a credits Token Plan on rolling 5-hour and 7-day windows; monthly cap removed.
Vendor’s framing: “the request-based Coding Plan has been fully upgraded to a Credits-based Token Plan”
Source: Digital Applied ledger (row 37) · Announced: 2026-07-19, official Qwen account on X · Plans: Alibaba Cloud Coding Plan ProCursor doubles included usage of Cursor models
Baseline Cursor Models pool (no figure published).2x that included usage.Details and source
Included usage in the Cursor Models pool (Auto, Composer, Cursor Grok) doubled on Pro, Pro+, Ultra and Teams; the third-party ‘Other Models’ dollar pool was not doubled.
Vendor’s framing: “we’ve doubled the included usage of Cursor models on all plans”
Source: Digital Applied ledger (row 20; original report: Startup Fortune) · Announced: 2026-07-16, Lee Robinson on X; restated by @cursor_ai 2026-07-21 · Plans: Cursor Pro (legacy)Claude Code +50% weekly boost extended to Jul 19
+50% weekly, ending 2026-07-13.+50% weekly, ending 2026-07-19.Details and source
First extension of the +50% weekly promotion; the ceiling did not move.
Vendor’s framing: Framed as keeping the increase
Source: Digital Applied ledger (row 6; original report: apidog) · Announced: On or about 2026-07-13 (exact day not established) · Plans: Claude Pro, Claude Max 5x, Claude Max 20xCodex five-hour limit temporarily removed (launch week)
Five-hour limit enforced alongside weekly limit.No five-hour limit (temporary).Details and source
During GPT-5.6 Sol launch week the rolling five-hour limit was removed for Plus, Business and Pro; weekly limits unchanged.
Vendor’s framing: A temporary removal during launch week
Source: Digital Applied ledger (row 17; original report: Morph) · Announced: 2026-07-12, Tibo Sottiaux on X · Plans: ChatGPT Plus (Codex), ChatGPT Pro 100, ChatGPT Pro 200Gemini CLI daily quota replaced by Antigravity weekly cap
Gemini CLI: 1,000 requests per day.Weekly compute-based cap; no absolute figure published.Details and source
Gemini CLI and Code Assist IDE extensions stop serving free/Pro/Ultra tiers; the Antigravity CLI meters a weekly compute cap with five-hour refreshes for Pro and Ultra.
Vendor’s framing: Migration to a successor product; quota change not headlined
Source: Digital Applied ledger (row 29) · Announced: 2026-05-19, Google I/O · Plans: Google AI Pro (Antigravity), Google AI Ultra 5x / 20x (Antigravity)Codex adds banked rate-limit resets
No user-controlled way to clear a rate limit.One banked reset per user at launch.Details and source
Rate-limit reset banking added: one free reset per user at launch, redeemable against a hit limit.
Vendor’s framing: A new capability
Source: Digital Applied ledger (row 16; original report: Ofox) · Announced: 2026-06-11, Codex changelog · Plans: ChatGPT Plus (Codex), ChatGPT Pro 100, ChatGPT Pro 200ChatGPT Pro $100: Codex usage drops from 10x to 5x Plus
10x ChatGPT Plus Codex usage (launch promo).5x ChatGPT Plus Codex usage.Details and source
The launch promotion (10x Plus Codex usage through May 31) ended; the standard 5x multiplier applies.
Vendor’s framing: Scheduled end of a launch promotion
Source: The Next Web · Announced: 2026-04-09 (end date stated at launch) · Plans: ChatGPT Pro 100Copilot switches to usage-based billing (AI Credits)
Premium request units × model multiplier.GitHub AI Credits (1 credit = $0.01) based on token consumption.Details and source
All Copilot plans now bill in GitHub AI Credits metered on input, output and cached tokens, replacing premium-request units.
Vendor’s framing: “all GitHub Copilot plans will transition to usage-based billing”
Source: GitHub Changelog · Announced: 2026-04-27 blog; live 2026-06-01 changelog · Plans: GitHub Copilot Pro, GitHub Copilot Pro+, GitHub Copilot Max, GitHub Copilot Business / EnterpriseCopilot Pro/Pro+ get a ‘flex allotment’ on top of base credits
Included usage equal to subscription price in credits.Pro 1,500 / Pro+ 7,000 / Max 20,000 credits per month.Details and source
Included usage became base + flex: Pro $10+$5=$15, Pro+ $39+$31=$70, Max $100+$100=$200. Base credits ‘never change’; flex ‘will vary over time’.
Vendor’s framing: “Base credits: matched 1:1 with your subscription price. These never change. Flex allotment: variable additional usage on top of your base.”
Source: GitHub Blog · Announced: 2026-05-12, GitHub blog · Plans: GitHub Copilot Pro, GitHub Copilot Pro+, GitHub Copilot MaxCopilot annual plans: model multipliers increased
Pre-June model multipliers.Higher multipliers — same allowance drains faster.Details and source
Annual Pro/Pro+ stay on premium-request pricing until expiry, but model multipliers increased for annual subscribers only; at expiry they move to Copilot Free unless converted.
Vendor’s framing: “Model multipliers will increase on June 1 (see table) for annual plan subscribers only.”
Source: GitHub Blog · Announced: 2026-04-27, GitHub blog · Plans: GitHub Copilot Pro, GitHub Copilot Pro+Antigravity: new $100 Ultra (5x Pro), top Ultra cut to $200 (20x Pro)
Pro $20 (entry quota) and Ultra $250 (highest quota).Ultra $100 = 5x Pro; Ultra $200 = 20x Pro; base Gemini entitlements adjusted upward.Details and source
Google standardised Antigravity multipliers: $100 Ultra = 5x Pro tokens, $200 Ultra = 20x Pro; AI credits removed from base plans and used only for overage.
Vendor’s framing: “we have adjusted the base plan entitlements upward for Gemini models on Antigravity”
Source: Google Antigravity Blog · Announced: 2026-05-19, Antigravity blog · Plans: Google AI Pro (Antigravity), Google AI Ultra 5x / 20x (Antigravity)Claude Code weekly limits +50% (temporary, to Jul 13)
Standard weekly limit (index 100).Weekly limit +50% (index 150), through 2026-07-13.Details and source
Weekly limits raised 50% for Pro, Max, Team and seat-based Enterprise; five-hour limits unaffected.
Vendor’s framing: “Claude Code weekly limits are increasing 50%, now through July 13”
Source: ScriptXeno · Announced: 2026-05-13, @ClaudeDevs on X · Plans: Claude Pro, Claude Max 5x, Claude Max 20xClaude Code peak-hour reduction removed
Reduced quota during weekday peak hours (since ~Mar 23).No peak-hour reduction.Details and source
The March peak-hour limit reduction was removed for Pro and Max accounts.
Vendor’s framing: “we’re removing the peak hours limit reduction on Claude Code for Pro and Max accounts”
Source: Anthropic · Announced: 2026-05-06, Anthropic news post · Plans: Claude Pro, Claude Max 5x, Claude Max 20xClaude Code five-hour limits doubled (permanent)
Baseline five-hour rate limit (no absolute figure published).2x baseline, permanent.Details and source
Anthropic doubled Claude Code’s five-hour rate limits for Pro, Max, Team and seat-based Enterprise, effective the same day.
Vendor’s framing: “we’re doubling Claude Code’s five-hour rate limits for Pro, Max, Team, and seat-based Enterprise plans”
Source: Anthropic · Announced: 2026-05-06, Anthropic news post · Plans: Claude Pro, Claude Max 5x, Claude Max 20xZ.ai cancels auto-renew on Legacy V1 plans
Auto-renewing V1 subscription.V1 runs to expiry with no renewal path.Details and source
Auto-renewal cancelled on eligible V1 plans; compensation of two complimentary months plus a 50% migration discount.
Vendor’s framing: Compensated migration
Source: Digital Applied ledger (row 32; original: Z.ai docs) · Announced: 2026-04-21 notice · Plans: Z.ai GLM Coding Plan (Lite / Pro / Max)Token-based Codex rate card extended to Enterprise, Edu and Gov
Per-message credit rate card.Token-based credit rate card.Details and source
The per-token credit rate card was extended to remaining existing Enterprise, Edu, Health, Gov and Teachers plans.
Vendor’s framing: Same as the April 2 change
Source: Digital Applied ledger (row 14; original: OpenAI Help Center Codex rate card) · Announced: On or before 2026-04-23, OpenAI Help CenterZ.ai: Legacy Plan V1 no longer supported going forward
V1 terms (no weekly cap) for the life of the subscription.Legacy plans phased out.Details and source
Z.ai announced legacy plans ‘will no longer remain as a supported subscription option going forward’.
Vendor’s framing: Phase-out with compensation
Source: Digital Applied ledger (row 31; original: Z.ai docs) · Announced: 2026-04-21, Legacy Plan Migration Notice · Plans: Z.ai GLM Coding Plan (Lite / Pro / Max)GitHub Copilot tightens individual-plan limits
Premium-request limits as previously published.Tighter limits; Pro+ more than 5x Pro.Details and source
Tighter usage limits on Pro and Pro+ (Pro+ >5x Pro), Opus removed from Pro, new Pro/Pro+/Student sign-ups paused; refunds offered until May 20.
Vendor’s framing: “Tighter usage limits for individual plans”
Source: GitHub Changelog · Announced: 2026-04-20, GitHub changelog · Plans: GitHub Copilot Pro, GitHub Copilot Pro+Alibaba ends Coding Plan Lite renewals and upgrades
Lite renewable and upgradeable.Lite runs to the end of the current term.Details and source
As of Apr 13 18:00 UTC+8, renewals and upgrades for Lite are discontinued.
Vendor’s framing: “Due to a product upgrade”
Source: Alibaba Cloud Model Studio docs · Announced: Alibaba Cloud announcement · Plans: Alibaba Cloud Coding Plan ProChatGPT Plus Codex allocation rebalanced toward steady daily use
Plus allowed longer burst sessions.Plus rebalanced toward steadier daily usage.Details and source
Announced with the new $100 Pro tier: Plus shifted toward steadier day-to-day usage rather than longer burst sessions.
Vendor’s framing: Presented as part of a tier launch, not as a reduction
Source: The Next Web · Announced: 2026-04-09 · Plans: ChatGPT Plus (Codex)Codex credits move from per-message to per-token pricing
Credits charged per message or per PR as approximate averages.Credits charged per 1M input, cached input and output tokens.Details and source
Codex credits for Plus, Pro, Business and new Enterprise switched to a token-based rate card (input, cached input, output). No conversion published.
Vendor’s framing: “we updated Codex pricing to align with API token usage, instead of per-message pricing”
Source: Digital Applied ledger (row 12; original: OpenAI Help Center Codex rate card) · Announced: On or before 2026-04-02, OpenAI Help Center · Plans: ChatGPT Plus (Codex), ChatGPT Pro 100, ChatGPT Pro 200Claude Code: faster five-hour burn at weekday peak hours
Uniform five-hour session burn rate.Faster burn during weekday peak hours.Details and source
Five-hour session limits drained faster on weekdays 05:00–11:00 PT; Anthropic estimated ~7% of users would newly hit limits. Weekly limits unchanged.
Vendor’s framing: “to manage growing demand for Claude we’re adjusting our 5 hour session limits”
Source: MacRumors · Announced: Not announced in advance; confirmed 2026-03-26 by Anthropic staff · Plans: Claude Pro, Claude Max 5x, Claude Max 20xAlibaba Coding Plan Lite closed to new subscribers
Lite available to new subscribers.Lite closed to new subscribers.Details and source
As of Mar 20 00:00 UTC+8 the Lite plan is no longer available for new subscriptions.
Vendor’s framing: “Due to a product upgrade”
Source: Alibaba Cloud Model Studio docs · Announced: Alibaba Cloud announcement · Plans: Alibaba Cloud Coding Plan ProWindsurf drops credits for daily + weekly quotas
Monthly credit pool (Pro: 500 credits/month).Daily and weekly quotas; extra usage at API pricing.Details and source
Credit pools replaced by usage allowances that refresh daily and weekly; overage at API pricing; existing paid subscribers got a free extra week.
Vendor’s framing: “No more credits… For the majority of users, this quota will be enough to fully cover all agent usage.”
Source: Windsurf (Cognition) blog · Announced: 2026-03-18, Windsurf/Cognition blog · Plans: Windsurf ProClaude: 2x five-hour limits off-peak (temporary)
Standard five-hour limits at all hours.2x the five-hour limit off-peak; ran to Mar 27 or 28 (sources conflict).Details and source
Temporary promo doubled five-hour limits outside weekday 08:00–14:00 ET and all weekend; bonus usage did not count toward weekly caps.
Vendor’s framing: “a small thank you to everyone using Claude”
Source: Digital Applied ledger (row 1; original report: WinBuzzer) · Announced: 2026-03-12/13 via @claudeai on X · Plans: Claude Pro, Claude Max 5x, Claude Max 20xGLM Coding Plan quietly adds a weekly cap
No weekly cap.~80/400/1,600 prompts per 5h and ~400/2,000/8,000 per week (Lite/Pro/Max).Details and source
An unannounced docs edit after the Feb 12 grandfathering deadline added weekly ceilings (~400/2,000/8,000 prompts) alongside five-hour allowances (~80/400/1,600).
Vendor’s framing: No framing — not characterised by the vendor
Source: Z.ai docs · Announced: Never announced; docs edit 2026-02-13 to 2026-02-16 · Plans: Z.ai GLM Coding Plan (Lite / Pro / Max)