[BUG] Claude Code Usage for Max Plan hitting limits extremely fast

[ggrillone]

Preflight Checklist

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

What's Wrong?

I have a Max subscription plan and all of a sudden hit the 5 hour usage limit in under 30 minute (Opus 4.6 high). When literally 2 days ago I was never hitting them in the 5 hour window and would have 2-3 hour long sessions of agents just running.

I just updated to the latest claude code (2.1.81) and then it started happening. I don't recall what the previous version was, I want to say the last time I updated claude code was when the mobile app remote feature was announced

Something is broken.

After the most recent hourly reset, I literally typed 1 message and it called a python script 2x and it made my hourly usage jump from 0 to 24% that makes absolutely not sense (I am on a Max Plan). Whatever is calculating usage right now is completely busted. I cannot get anything meaningful done.

Terminal output:

 we had to stop because of usage limits, which are now reset. review where we left off and then continue                   

⏺ Bash(python3 ./scripts/craft-tools.py state read \
        --task-dir ./tasks/LOCAL-032 --operation current-step 2>&1)
  ⎿  {
       "ok": true,
       "step": {
     … +6 lines (ctrl+o to expand)

⏺ We left off at spec-critique in-progress. The last round of reviews completed successfully:
  - Spec Critic R2: APPROVED — 0 blockers
  - Architect R2: No blocking issues, all 7 revisions verified

  But the workflow state still shows spec-critique as in-progress because we hit the usage limit before updating it and
  proceeding to Gate 1. Let me mark it complete and proceed.

⏺ Bash(python3 ./scripts/craft-tools.py state write \
        --task-dir ./tasks/LOCAL-032 --operation set-step-status \…)
  ⎿  {
       "ok": true,
       "step": "spec-critique",
     … +2 lines (ctrl+o to expand)

What Should Happen?

My usage (on a Max Plan) should not be hit so quickly. It never did this just 2 days ago.

Error Messages/Logs

Steps to Reproduce

1. Enter a claude session via claude
2. Prompt the agent to perform a task like read a file, write, call a script

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

No response

Claude Code Version

2.1.81

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

No response

[github-actions]

Found 3 possible duplicate issues:

1. [BUG] Max x20 plan — 5-hour quota consumed in ~10 prompts (started March 14, 2026) #34410
2. [BUG] Hitting much faster usage limits than before #28537
3. [BUG] Regression v2.1.74: ~3% quota consumed per 20k token session — Max20 plan #33752

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

[ggrillone]

It is no longer possible to complete meaningful work with Claude Code. Please fix this. @claude

[ggrillone]

So far it seems better, but only after I started a new Claude Code chat session. My new session is using the same model (Opus 4.6) and same reasoning effort (high) as my previous session that was hitting hourly limits way too quickly.

It does feel like maybe the bug for the usage skyrocketing is more related to continuing a Claude Code session after updating Claude Code.

Before I started this new session, I prompted the old session to create a primer prompt with all the context the new session needed. It read 3 markdown files (totaling about 1,500 lines) and then output a 50 line prompt. And that cost me 55% of my hourly usage on my Max Plan. Insane.

At this point I've hit >50% weekly limit because of this bug.

But so far so good with the new chat session, the usage is going up at reasonable rate and I'm actually able to complete my tasks.

I hope this information helps.

[GitOviSAn]

similar issue. woke up this morning, had a terminal loaded already with a bit of context, did like 2 prompts and BOOM 100% and the reset just happened less than 45 mins ago there is no way i used it all!!
seems to be more related to the APP claude Code rather than the terminal (just started using app yesterday and this happened this morning)

UPDATE VERY UPSETTING: MY All models is ALREADY AT 16% used!!!!!! from this same issue i haven't even been able to use it this morning and i'm at 16% all usage!!!?!?!?!?!? its crazy how frustrating this is...

I very much hope they refund this or fix it. i had a lot of tasks this morning to achieve that are now disrupted. appreciate you guys keep up the good work.

[paulokinho]

Same here, I'm subscribed to 20x but today it burned the 5 hour limit in exact 85 minutes, my work load was exact the same as any other day.

[GitOviSAn]

Same here, I'm subscribed to 20x but today it burned the 5 hour limit in exact 85 minutes, my work load was exact the same as any other day.

also on Max. i didn't do anything different either... this is definitely on them. i just keep refreshing the usage page praying its just a bug... what to do with 5 hours of no Claude.... HELP!

[jgseddon]

Having this issue too! my limit reset and within 2 minutes it maxed out on pro plan! not even a big request - something is definitely wrong here...

[GitOviSAn]

URGENT! DO NOT MAKE CLAUDE DO ANY WORK! WILL TAKE MOST IF NOT ALL USAGE IN ONE GO FOR THE 5HR WINDOW AND UP TO 20% IN WEEKLY! THIS IS A LOSS OF $20's payment IN A SINGLE PROMPT!

[tommyxps]

Haiving the same issue, I have encountered the blockage twice in a row toady in 5 + 5 hours window on $100 Max plan. This never happened before, please fix....

[ThatDragonOverThere]

Confirming on v2.1.81, Max plan, Windows CLI, single conversation.

Woke up this morning, started one conversation, typed for ~20 minutes, hit the 5-hour rate limit. This has never happened before — I routinely run multi-hour sessions with heavy sub-agent usage and don't come close to the 5-hour limit.

This was ONE conversation. Not multiple windows, not parallel agents. Normal usage pattern that has worked fine for months.

Something changed server-side. This isn't a client version issue.

[chsrisoph]

Same issue, i'm using MAX significant reach limit x5 times - feel like when use Pro with the same task.