[BUG] [Max plan] Weekly usage limit reset day shifting forward each week (Wed → Thu → Fri)

Resolved 💬 3 comments Opened Apr 30, 2026 by YigitYerlitas Closed May 31, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

Summary

On the Claude Max plan, my weekly usage limit reset day is
shifting forward by ~1 day each week instead of staying on a
fixed weekday or rolling cleanly on a 168-hour cadence.

Timeline

  • Initial subscription: Weekly limit reset on Wednesday
  • Last week: Reset shifted to Thursday
  • This week (2026-04-30): Usage page shows next reset on Friday

Net shift: +2 days over 2 weeks.

Expected behavior

Per the Help Center article "How do usage and length limits
work?", the weekly limit should follow a rolling 7-day window
from the first session of the week. In practice, this should
either:

  • Stay on the same weekday if my usage pattern is consistent, or
  • Roll cleanly on a 168-hour cadence

Actual behavior

The reset day shifts forward by roughly one calendar day per
week, which suggests the rolling window is being recalculated
with an additional offset (possibly a buffer or a recalculation
trigger when the limit is hit).

Related issues

  • #29680 — earlier report of weekday shift on Max plan (Feb 2026)
  • #52921 — 24-hour cycle bug on Max 20x (Apr 2026)

Environment

  • Plan: Claude Max
  • Account region: Turkey
  • Observed via: claude.ai → Settings → Usage
  • Date observed: 2026-04-30

Screenshot

<img width="1707" height="951" alt="Image" src="https://github.com/user-attachments/assets/f0686b02-8645-4263-a199-46925aa07082" />

Question for the team

Is this the documented rolling-window behavior, or is my
account affected by the same calculation bug as #29680/#52921?
If the former, can the docs be updated to clarify that the
reset day will progressively shift?

What Should Happen?

Under the documented 168-hour rolling window, my weekly reset
day should either stay on the same weekday week after week (if
my first-prompt time is consistent) or shift backward (if I
start earlier in the cycle the following week). It should never
shift forward, because forward shift requires the window to be
longer than 168 hours.

Concretely, expected behavior:

  • If my first prompt of week 1 was Wednesday 14:00, my reset

should be next Wednesday 14:00.

  • If my next first prompt happens at Wednesday 14:01 or later,

reset stays on Wednesday.

  • If my next first prompt happens earlier (e.g., Tuesday),

reset shifts backward to Tuesday.

  • A reset shift from Wednesday → Thursday → Friday over three

weeks should not be possible.

Either:

  1. The reset calculation should match the documented 168-hour

rolling window (so the day stops drifting forward), OR

  1. The Help Center documentation should be updated to describe

the actual mechanism — including any buffer, recalculation
trigger, or condition under which the window exceeds 168
hours.

Whichever is the correct fix, the user-facing behavior and the
documented behavior should match.

Error Messages/Logs

Steps to Reproduce

  1. Subscribe to Claude Max plan.
  2. Note the weekday on which your weekly limit first resets

(in my case: Wednesday).

  1. Continue normal usage for 2-3 weeks.
  2. Each week, check claude.ai → Settings → Usage and record

the displayed "next reset" weekday.

  1. Observed: reset day shifts forward by ~1 day each week

(Wed → Thu → Fri).

  1. Expected: reset day stays the same or shifts backward,

per the documented 168-hour rolling window.

Note: I do not have screenshots from previous weeks, only
from today (2026-04-30). However, Anthropic's backend should
have my reset history.

Claude Model

Not sure / Multiple models

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

1.4758.0

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Terminal.app (macOS)

Additional Information

Support acknowledgment
Fin Conversation ID: 215474123652090 (Anthropic engineering
can pull the full transcript from this ID).

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗