RESET OpenAI resets every paid Codex and ChatGPT meterWATCH Codex lead’s 28-day limits pledge hits Day 8RUMOUR Grok, Cursor and X Premium may merge. UnconfirmedREVIEW OpenAI polls Pro users on how limits should workCAPPED Claude Max: Fable capped at 50% of the weekly limitFINE PRINT Claude Max has three meters, not oneFINE PRINT Claude Max API credits don’t work in Claude CodeFINE PRINT ChatGPT Plus Codex: model choice swings limits up to 100×FINE PRINT SuperGrok Heavy is a usage grant, not a Cursor planFINE PRINT Codex Fast mode burns limits 2.5×, Ultrafast 8×NERFED New ChatGPT Pro $200 subscribers get less, same priceCUT Old ChatGPT Pro $200 allowance ends 29 OctNEW Grok Bot gets its own weekly pool in Cursor Pro+
Got CappedTue, 13 Oct 2026 · 1 reports logged
Pooled, self-reported AI plan limits

Got capped?

Log it in 30 seconds.

Vendors show you a percent. Nobody shows what the percent is worth. Drop a screenshot of your usage page, or tap three answers, and we pool it into each plan's real capacity, week by week, so you can see when a plan gets cut.

1reports this week
14plans tracked
13sourced changes
16days to: ChatGPT Pro $200: grandfathered allowances end

Got capped? Log it.

Takes 30 seconds
Drop a screenshot of your usage pageor tap to choose one. You can also paste it.
Your plan
What ran out?
When?

The board

Median capacity this week · self-reported
PlanVs last weekCapacitynChange
Claude Pro$20EFNO READING—log the firstn=0—
Claude Max 5x$100EFNO READING—log the firstn=0—
Claude Max 20x$200EFNO READING—log the firstn=01 hit logged—
ChatGPT Plus$20EFNO READING—log the firstn=0—
Cursor Pro$20EFNO READING—log the firstn=0—
Cursor Ultra$200EFNO READING—log the firstn=0—
SuperGrok$30EFNO READING—log the firstn=0—
Copilot Pro$10EFNO READING—log the firstn=0—
Copilot Pro+$39EFNO READING—log the firstn=0—

Capacity = tokens or messages used ÷ the percent of the limit the vendor showed as used. No percent means you hit the limit. Self-reported and unverified; a median shows at 5 reports in a week and extreme values are trimmed. The gauge reads this week's median against last week's, so it needs both weeks.

Cuts this week

Median capacity vs last week · both weeks n≥5

No week-over-week moves yet. A plan shows here once it has 5 reports with usage figures this week and last week. Until then, the sourced changes below are the record.

What changed

RSS
13 Oct
RESET

OpenAI resets every paid Codex and ChatGPT meter

Global usage reset for paid ChatGPT and Codex accounts, announced and confirmed by Codex lead Tibo. Weekly meters went back to full. It was a one-off, not a new schedule.

x.com · ChatGPT Plus

13 Oct
WATCH

Codex lead’s 28-day limits pledge hits Day 8

Tibo's 28-day pledge reaches Day 8 ("Reset"). He is posting a daily update on Codex limits. Expect more limit changes this month. Check back.

x.com

13 Oct
RUMOUR

Grok, Cursor and X Premium may merge. Unconfirmed

Posts on X say Grok, Cursor and X Premium may merge into one plan. Not confirmed by xAI or Cursor. Wait for an announcement before switching.

x.com · unconfirmed

11 Oct
REVIEW

OpenAI polls Pro users on how limits should work

Tibo polls Pro users on how they'd like limits to work. 5M views, 5,600 replies. Pro limits are openly under review.

x.com · ChatGPT Pro $200

6 Oct
CAPPED

Claude Max: Fable capped at 50% of the weekly limit

Fable 5 and 5.1 are included on Max 5x and 20x but can use only 50% of the weekly limit. On Pro, Fable is billed as usage credits. Heavy Fable use runs out halfway through the week.

claude.com · Claude Max 20x

6 Oct
FINE PRINT

Claude Max has three meters, not one

Max has a separate weekly limit for Fable, on top of the 5-hour session and the general weekly limit. All three are shared between Claude and Claude Code. Three meters to watch, not one.

support.claude.com · Claude Max 5x

6 Oct
FINE PRINT

Claude Max API credits don’t work in Claude Code

Max includes $100 or $200 a month in API credits. They can't be spent in Claude Code or the Claude apps. The credits don't extend your Claude Code limit.

support.claude.com · Claude Max 5x

6 Oct
FINE PRINT

ChatGPT Plus Codex: model choice swings limits up to 100×

Plus Codex estimates per 5 hours: GPT-6 Astra 5–45, GPT-6 Sol 15–150, GPT-6.1 Sol 15–160, Luna 350–3,000 local messages. Weekly limits may also apply. The model you pick changes your allowance by up to 100×.

tokenplans.dev · secondary source · ChatGPT Plus

5 Oct
FINE PRINT

SuperGrok Heavy is a usage grant, not a Cursor plan

Linking SuperGrok Heavy grants the highest weekly Grok Bot usage on a Cursor account. It's a usage grant, not a Cursor plan. Heavy doesn't add on top of a paid Cursor plan's own pool.

cursor.com · SuperGrok Heavy

2 Oct
FINE PRINT

Codex Fast mode burns limits 2.5×, Ultrafast 8×

Pro 500 is the only Pro tier with Astra Ultrafast. Fast mode uses the allowance 2.5× faster and Ultrafast 8× faster. Speed modes burn limits fast. Check what mode you're in.

tokenplans.dev · secondary source

29 Sept
NERFED

New ChatGPT Pro $200 subscribers get less, same price

Pro $200 reopens to new subscribers with a lower allowance than the old Pro 20x. Price unchanged. New $200 subscribers get less than existing ones.

developersdigest.tech · secondary source · ChatGPT Pro $200

29 Sept
CUT

Old ChatGPT Pro $200 allowance ends 29 Oct

Existing Pro $200 subscribers keep their old allowance until 29 October. Grandfathered users drop to the new allowance after that date.

tokenplans.dev · secondary source · ChatGPT Pro $200

21 Aug
NEW

Grok Bot gets its own weekly pool in Cursor Pro+

Grok Bot bundled into Cursor Pro+ with its own weekly usage pool, separate from Cursor's model pools. Grok Bot use doesn't eat your Cursor pool, but it has its own weekly cap.

tokenplans.dev · secondary source

The plans

Each card is shareable
n=0

Need 5 reports this week to show a number.

5-hour session + weekly. At least 5× Free per session. No message count.

n=0

Need 5 reports this week to show a number.

5-hour session + weekly + Fable weekly. 5× Pro per session. Weekly figure not published.

n=0

Need 5 reports this week to show a number.

5-hour session + weekly + Fable weekly. 20× Pro per session. Weekly figure not published.

n=0

Need 5 reports this week to show a number.

5-hour + weekly (Codex). Codex estimates per 5h: Astra 5–45, Sol 15–160, Luna 350–3,000 local messages.

n=0

Need 5 reports this week to show a number.

Weekly only. 5× Plus. No five-hour limit. Weekly figure not published.

n=0

Need 5 reports this week to show a number.

Weekly only. 20× Plus. New subscribers since 29 Sep get a lower allowance.

n=0

Need 5 reports this week to show a number.

Monthly $ pool. $20/mo API pool + Cursor models pool.

n=0

Need 5 reports this week to show a number.

Monthly $ pool. $400/mo API pool + Cursor models pool.

n=0

Need 5 reports this week to show a number.

Not published. "4× limits" versus Free.

Google AI Ultra$99.99 / $199.99
n=0

Need 5 reports this week to show a number.

Not published. Ultra 5x = 5× Pro; Ultra 20x = 20× Pro.

n=0

Need 5 reports this week to show a number.

Not published. No limit figure published. Includes weekly Grok Bot usage.

n=0

Need 5 reports this week to show a number.

Not published. No limit figure published. Highest linked Grok Bot weekly usage.

n=0

Need 5 reports this week to show a number.

Monthly credits. $15/mo AI credits. Completions unlimited.

n=0

Need 5 reports this week to show a number.

Monthly credits. $70/mo AI credits (4× Pro).

Follow a plan

One email when its limits change
Advertise here. One sponsor at a time, on this page and in the alerts.Sponsor weekly · $150Sponsor monthly · $500