Usage limits hit significantly faster since v2.1.53–v2.1.59 (Feb 25–26) — auto-memory & system prompt bloat

the fact it takes over 15 minutes and a 32000 token limit to make one small color change is absolutely ridiculous .

Found 3 possible duplicate issues:

1. https://github.com/anthropics/claude-code/issues/29169
2. https://github.com/anthropics/claude-code/issues/29175
3. https://github.com/anthropics/claude-code/issues/28537

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

👎

👎

Same issue here on the Pro plan. Normally don't have a problem with usage, but I hit the 32k token limit for the first time today on a comparatively simple query.

Which version is the most stable version? I have also encountered this situation now. The number of token input has increased sharply, resulting in more tokens consumption.

Hi. This is fixed in v2.1.62. Please run claude update. We've also reset rate limits: https://x.com/trq212/status/2027232172810416493

Same issue here, Max plan (Feb 26)

I've been using Claude Code every day for months now. I know my usage patterns well. Sometimes I do burn through quota fast but that's always when I'm doing heavy complex work across multiple tasks.

Yesterday (Feb 26) around 19:00-20:00 UTC+0 I wasn't doing anything special at all. Session reset and within about 20 minutes my entire usage was gone. Completely locked out for the rest of the period.

Nothing changed on my side. Same workflows, same type of tasks. Months of daily use and I've never seen anything like this.

Some context on my setup:

• Max plan
• I had the $43 extra usage credits they offered a while back, already used those up. Luckily I had set a spending limit so I didn't get charged extra, but I was totally blocked from working
• The usage consumed was way out of proportion to what I was actually doing
• This started right around the Feb 25-26 update window

If this is the auto-memory and system prompt bloat issue, it makes sense. Invisible tokens getting injected into every single message would explain why even a light session drained everything so fast. But for paying Max users who rely on this tool for daily work, getting silently locked out with no explanation is really frustrating.

something is wrong here, specially after 26 FEB.
one prompt consumed 32k token and the session percentage became 23% ??? im in max plan. its not about auto memory or system prompt, something is wrong in the session tokens calculations !!!

I am still experiencing the same issue even after upgrading to v2.1.62.
It is consuming tokens massively and i am reaching limits of the Max Plan significantly faster, anyone can confirm if the issue is solved ?

It’s not. I’m still having the same exact issues on 2.1.62

We hit this exact issue — system prompt bloat + auto-memory eating tokens faster than expected. We built cozempic to deal with it. cozempic diagnose shows you the exact token breakdown so you can see how much is conversation vs system overhead vs stale tool results.

It also prunes the bloat that accelerates compaction — progress ticks, duplicate system-reminders, oversized tool outputs, stale file reads. Less noise = more useful context per token = slower limit consumption.

pip install cozempic
cozempic current --diagnose

Would love feedback if anyone tries it.

@shawnm-anthropic we need feedback on this from anthropic team as this has not been resolved and as paying customers it is quite upsetting

We are back at usage draining issue, today I am literally noticing my usage goes down almost instantly

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.