[BUG] CLAUDE CODE Hitting usage limits after barely over an hour of usage. (max plan), when previously I had it set to auto (opus till near limit)...worked all day, NEVER hit limits. WTF!?
Environment
- Platform (select one):
- [ ] Anthropic API
- [ ] AWS Bedrock
- [ ] Google Vertex AI
- [ ] Other: <!-- specify -->
- Claude CLI version: <!-- output of
claude --version--> CLAUDE CODE - Operating System: <!-- e.g. macOS 14.3, Windows 11, Ubuntu 22.04 -->
- Terminal: <!-- e.g. iTerm2, Terminal App --> CLAUDE CODE
Bug Description
I hit my usage limit with the max plan in less than an hour and a half. I've never ever had that happen before. Had it set to Auto...so I assume it was working with Opus till last '20%', It's lot like even used many messages that I sent. Aren't those the only one's counted? :
Steps to Reproduce
- <!-- First step -->
- <!-- Second step -->
- <!-- And so on... -->
Expected Behavior
Same as always, just use it til rate limits hit...but RARELY HIT IT!
Actual Behavior
<!-- What actually happened -->
Additional Context
<!-- Add any other context about the problem here, such as screenshots, logs, etc. -->
claude-code-analytics
1.9.0
real-time monitoring dashboard
last update: 7/17/2025, 1:50:19 PM
🌙
⭐
Star on GitHub
conversations:
122
sessions:
26
tokens:
3,723,759
projects:
3
storage:
237.96 MB
▼
Current Session - Pro Plan
Active
1h 13m
remaining
Messages
6/45
ℹ️
Session Time
1h 7m/2h 21m
from:
07/10/2025
to:
07/17/2025
refresh charts
📊 token usage over time
🎯 project activity distribution
🛠️ tool usage trends
⚡ tool activity summary
Total Tool Calls
5212
Unique Tools Used
13
Conversations with Tools
54/62
Avg Tools per Conversation
96.5
Tool Balance Score
74%
🏆
Most Used: Read
1247 calls (23.9%)
← back to conversations
conversation: 96e3a446...
CSV
export
refresh
conversation id
96e3a446-b89b-4141-b099-e3ff468eeaa6
project
Extention
messages
145
total tokens
7,313
model
claude-opus-4-20250514
service tier
standard
token details
200 in / 7,113 out
cache tokens
121,277 created / 2,706,023 read
file size
521.71 KB
created
7/16/2025, 9:52:10 PM
last modified
7/17/2025, 11:41:51 AM
conversation state
User typing...
15 Comments
same here
I'm sure it's everybody. I know it's happening to me. I feel cheated and lied to.
Same issue here - noticeable reduction in amount of work that can be done before hitting usage limits.
i just got a opus limit after about 20 minutes of coding. that feels dramatically reduced over what I was getting 7 - 10 days ago.
I mean...it feels like there's room here for a class action lawsuit. This type of rugpull is crazy!! But i'm sure whatever we signed to even use it says something about not being obligated to tell us how much we'll be able to use it even though it's advertised as... 'there are no limits' when you ask it where you stand on your current usage. IT'S BEYOND INFURIATING! Not to mention...Sonnet 4 seems like it's gotten INCREDIBLY MORE STUPID! It's basically useless now. WTF!
I mean...it feels like there's room here for a class action lawsuit. This
type of rugpull is crazy!! But i'm sure whatever we signed to even use it
says something about not being obligated to tell us how much we'll be able
to use it even though it's advertised as... 'there are no limits' when you
ask it where you stand on your current usage. IT'S BEYOND INFURIATING! Not
to mention...Sonnet 4 seems like it's gotten INCREDIBLY MORE STUPID! It's
basically useless now. WTF!
On Mon, Jul 21, 2025 at 10:12 AM Ben Joldersma @.***>
wrote:
--
Danny Jones
605-484-3913
@.***
Absolutely. Very obvious drop in any introspection before response. Now it codes without considering the context of how the code will run.
They have to be working on this. Claude seemed smarter yesterday. Found 60 lines of repetitive code and condensed it to 8. This and several similar actions yesterday seemed like there were improvements amongst the outages. It made almost no mistakes yesterday. Very encouraging
Using Claude sometimes feels like watching a glitchy movie in a bad theater - you pay for the ticket, the film freezes halfway through, no refunds, and by the time it resumes, you wish you'd brought popcorn and a debugger. Limits hit fast, and you're left wondering if writing the spec was easier than writing code manually and doign everything yourself from scratch. MCP boosts performance a little but mainly in hitting those limits faster. And switching to API or open alternatives? Let's just say... the sequel isn’t much better. I use automatic model selection, and try to keep claude up to date, so probably they tweak their internal prompts and if you endup using opus more frequently you quickly eat into limits.
Yes, this is becoming a money-grab bait-and-switch game. I used to be able to work all day without hitting the limit. Now I'm hitting the limit on the Max plan every couple hours. Really disgusting. These are the kinds of tactics that enrage customers, kills loyalty, and drives them to your competitors.
Now every day I'm looking to switch to better models with lower prices because of this BS.
Claude max. Same here hit the limit today in 1.5 hour. Did not do anything special, used context7, playwright mcp, but don't know how I managed to hit that when I didn't use it much.
I moved into Max yesterday. Have yet to hit limit. 14 hours today. Coding does still setem acking but better than other options. However the coding a month or so ago was amazing and then a short time after that back to not so amazing. I'm building an mcp server to enforce a detailed coding plan. All bugs, yeah right, worked out of it today. Looking forward to testing it tomorrow. Also made an mcp server that allows power shell access for cruising around file system, making edits, etc anything you would do in CMD/power shell. Also it allows it to SSH into my server and do work for me. THAT has been amazing. Made my life smooth like butter.
Found 3 possible duplicate issues:
This issue will be automatically closed as a duplicate in 3 days.
🤖 Generated with Claude Code
This issue has been automatically closed as a duplicate of #3650.
If this is incorrect, please re-open this issue or create a new one.
🤖 Generated with Claude Code
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.