[BUG] 5 hour limit reached after only 3 hours
Environment
- Platform (select one):
- [ ] Anthropic API
- [ ] AWS Bedrock
- [ ] Google Vertex AI
- [X ] Other: Claude Code 4 Sonnet
- Claude CLI version: <!-- output of
claude --version--> - Operating System: MacOS 14.7.1
- Terminal: <!-- e.g. iTerm2, Terminal App -->
Claude Code v1.0.89
L Session ID: 5966a933-6df3-4017-9468-bbf5f50bdf52
Working Directory
L /Users/iwolf/Repos/beascout
IDE Integration • /config
✔ Connected to Visual Studio Code extension
L Installed VS Code extension version 1.0.89 (server version: 1.0.57)
Account • /login
L Login Method: Claude Pro Account
L Organization: irawolf81@gmail.com's Organization
L Email: irawolf81@gmail.com
Model • /model
L Sonnet Sonnet 4 for daily use
Bug Description
At 10pm 22Aug2025 EST, I was notified that my limit was reached and that it would reset at 12am (midnight). I resumed my Claude Code session at midnight, accidentally closed my VS Code instance, reopened it and started a new Claude Code session. At 3am 23Aug2025, I was notified with "5-hour limit reached ∙ resets 5am". I was active for only 3 hours, yet my 5 hour limit was reached.
Steps to Reproduce
See Bug Description
Expected Behavior
I expected to be notified my 5 hour limit was reached at 5am after resuming session at midnight.
Actual Behavior
My 5 hour session was terminated 2 hours early.
Additional Context
Did the 5 hour time incorrectly start when my prior session was halted instead of when it was resumed?
<img width="990" height="375" alt="Image" src="https://github.com/user-attachments/assets/0ceeaed2-4108-4e7b-ba29-85f6c95b8db5" />
10 Comments
Found 2 possible duplicate issues:
This issue will be automatically closed as a duplicate in 3 days.
🤖 Generated with Claude Code
What it means is the token limit has been reached, and you have to wait for the next 5-hour cycle, when it will reset.
It doesn't mean you can use it for 5 hours or something like that, but that for every 5 hours you have a certain amount of tokens to spend. Besides the 5-hour window, there's also a weekly window recently introduced, but I don't know much about that yet.
Here's more info:
<img width="837" height="876" alt="Image" src="https://github.com/user-attachments/assets/73e7fb81-3540-4cce-a94e-f62c2701aac6" />
https://portkey.ai/blog/claude-code-limits/
With Opus model, it's not that hard to reach that limit, but if you switch to Sonnet, you'll be able to run a few Claude Code instances in parallel without hitting the token limit.
I am actually grateful for the follow 5 hour limit now. It has kicked me off my computer at 3am for two consecutive nights, thus forcing me to sleep.
Thank you for the useful replies. They have been informative.
this has been happening to me more recently. i am no where near the 5 hours and i also run things like claude-code-monitor and it shows i am no where near the token usage. prior to this week the monitor was pretty spot on for the limits. now the predictions are no where near what actually happens. take tonight. i started at 8 it stopped me at 10
📊 Session-Based Dynamic Limits
Based on your historical usage patterns when hitting limits (P90)
────────────────────────────────────────────────────────────
💰 Cost Usage: 🟡 [██████████████████████████████████░░░░░░░░░░░░░░░░] 68.0% $8.93 / $13.13
📊 Token Usage: 🟢 [████████████████████████░░░░░░░░░░░░░░░░░░░░░░░░░░] 48.6% 22,558 / 46,417
📨 Messages Usage: 🟡 [█████████████████████████████░░░░░░░░░░░░░░░░░░░░░] 58.6% 178 / 304
────────────────────────────────────────────────────────────
⏱️ Time to Reset: 🟡 [██████████████████████████████████░░░░░░░░░░░░░░░░] 1h 34m
🤖 Model Distribution: 🤖 [██████████████████████████████████████████████████] Sonnet 100.0%
────────────────────────────────────────────────────────────
🔥 Burn Rate: 109.9 tokens/min ➡️
💲 Cost Rate: $0.0435 $/min
🔮 Predictions:
Tokens will run out: 01:01
Limit resets at: 01:00
is there a better tool to use or something?
Yesterday, my Sonnet 4 sessions ended after two hours. Today, my 5-hour limit ended after just 70 minutes of working, which includes time for offline analysis. This experience is starting to feel like arcade games where the closer you get to completing a level, the more tokens you have to feed the game. It feels like there is a congestion surcharge for using system during normal business hours; however, it's more likely that my context cannot be further compressed.
For a while, I was working very well with a Claude session the seemed to understand me perfectly. It was a great pair programming partner! Then my VS Code window crashed and I got a new Claude session. Even through I have a detailed description of my professional background, development philosophies, and AI interaction preferences, as well as a maintaining project design details and state, all of which I reloaded into the new session, Claude still felt like a stranger to me. It no longer could anticipate nor complete my thinking as well as it did in the prior session. The interactions improved after a while but not to the point that they were before.
And now my token limit is being reached sooner and sooner, slowing down previously amazingly fast progress.
I have the exact same issue happening to me where the session reach its limit even though I am nowhere near the token nor the time limit:
📊 Session-Based Dynamic Limits
Based on your historical usage patterns when hitting limits (P90)
────────────────────────────────────────────────────────────
💰 Cost Usage: 🟡 [█████████████████████████████████████░░░░░░░░░░░░░] 75.0% $8.33 / $11.11
📊 Token Usage: 🟢 [█████████████████░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░] 34.9% 18,071 / 51,734
📨 Messages Usage: 🟢 [███████████████████████░░░░░░░░░░░░░░░░░░░░░░░░░░░] 47.7% 198 / 415
────────────────────────────────────────────────────────────
⏱️ Time to Reset: 🟡 [████████████████████████████████░░░░░░░░░░░░░░░░░░] 1h 44m
🤖 Model Distribution: 🤖 [██████████████████████████████████████████████████] Sonnet 100.0%
────────────────────────────────────────────────────────────
🔥 Burn Rate: 92.5 tokens/min ➡️
💲 Cost Rate: $0.0427 $/min
🔮 Predictions:
Tokens will run out: 12:20 PM
Limit resets at: 1:00 PM
⏰ Cost limit will be exceeded before reset!
Happened again 1h30 after the reset...
Such a big misunderstanding what it actually is...
It actually is very clear and even documented if you would read it.
I cant remember the exact numbers but you get X message per session. A session is defined by 5 hours from the first prompt.
If you use more message than your tier/session allows within that 5 hour session? You get cut off and use until the time is reset.
Its also hard capped to 50 of those sessions per month and in about 22 hours? They are also enforcing a weekly cap in combination to the monthly cap while keeping 50 sessions... Its so lame...
I get the 5-Hour BS right after entered the CLI. I hadn't used CC for about a month. The subscription is paid and still active.
Very confused and disappointed.
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.