[BUG] Weekly usage stuck at 0% on claude.ai and CLI across two Pro accounts despite heavy usage (recurring, 3rd time this month)

Open 💬 0 comments Opened Jul 10, 2026 by 4keles

<img width="1414" height="634" alt="Image" src="https://github.com/user-attachments/assets/5e093b99-0151-4d9e-8899-3a3ca493bf2f" />
<img width="736" height="334" alt="Image" src="https://github.com/user-attachments/assets/b1309143-41d4-448c-8983-ecbed18cc25a" />
<img width="1440" height="917" alt="Image" src="https://github.com/user-attachments/assets/e4871c50-e89b-4272-9ecf-1b80ae1d95bd" />
<img width="770" height="287" alt="Image" src="https://github.com/user-attachments/assets/bf68f3ad-0743-48d0-b977-daa4625ee650" />
<img width="1596" height="955" alt="Image" src="https://github.com/user-attachments/assets/a0e824cd-d9a2-4889-8ce5-5f2de760e2b6" />

Summary

Weekly usage ("Current week (all models)") repeatedly shows 0% used despite substantial actual usage that should already be ~40-50% at that point in the week. Reproduced independently on two separate Pro accounts (same Google account, two different emails/orgs), and on both the CLI (/usage) and the claude.ai web app (Settings → Usage) at the same time. This is the 3rd occurrence this month — pattern is: a period of heavy usage, then some time later (often first noticed on opening the web app), weekly usage shows 0%.

Environment

  • CLI version: 2.1.206 (account 1), 2.1.205 (account 2)
  • Plan: Pro (both accounts)
  • Login method: claude.ai OAuth
  • OS: Linux

Evidence

Account 1/usage (CLI) and claude.ai Settings → Usage both showed at the same moment:

  • Current session: 4% used
  • Current week (all models): 0% used (resets Jul 12, 2am)
  • claude.ai side panel: "Last updated: just now" — so this wasn't a stale/cached client value, the server itself reported 0%

Account 2/usage (CLI):

  • Current session: 42% used, total session cost $352.25, 14h43m API duration
  • Current week (all models): 5% used (resets Jul 13)
  • Note: this 5% appeared only because a new agent had just started running in parallel at the moment of the screenshot — i.e., weekly usage began incrementing from a fresh new request, not from the substantial usage that should already have accumulated earlier in the week.

Why this looks like a server-side bug, not a display/cache issue

  • Reproduced identically on claude.ai web (not just the CLI), which rules out a purely local/CLI cache explanation
  • claude.ai explicitly showed "Last updated: just now" while still reporting 0%
  • The counter does respond to fresh activity (account 2 ticked up to 5% the moment new usage occurred), so the increment mechanism itself works — but the pre-existing cumulative usage for the week appears to have been silently dropped/desynced
  • Recurring: this is the 3rd time this month the reporter has seen this exact pattern

Impact / still verifying

Not yet confirmed whether this is purely a display bug or whether it also affects the actual enforced weekly limit (i.e., whether more real usage is being granted than the plan allows, or a block will incorrectly trigger later this week). Reporter is monitoring account 2 for the rest of this week's cycle and will follow up with results.

Related issues

  • #73751 (weekly % shows 0% while /usage reflects real usage)
  • #75222 (usage shows 0% but access blocked)

Screenshots

Will attach CLI /usage and /status output plus claude.ai Usage settings panel for both accounts in a follow-up comment (image upload isn't available via the API/CLI used to file this issue).

View original on GitHub ↗