[BUG] Claude Code freezes during thinking or at a random token while still consuming usage limits
What's Wrong?
Claude Code sometimes freezes halfway through a response - either during the extended thinking phase or on a random token - and completely stops generating output. Despite no visible activity, it continues to consume tokens and user limits. The only workaround is CTRL+C and resending the prompt.
Environment
System: Windows 11
Installation method: native installer (irm https://claude.ai/install.ps1 | iex)
Model: claude-sonnet-4-6 (default)
What Should Happen?
Expected behavior
Claude Code should either complete the response or return a timeout/error and stop consuming limits. In its current state, this can be perceived as misleading for users. Claude experiences failures relatively often, and users receive no compensation, which may be considered an unfair practice.
Actual behavior
The process stalls, does nothing, and tokens/limits continue to be consumed. No error message is shown.
Steps to Reproduce
- Run Claude Code on any project (e.g. C# code analysis)
- Send any prompt
- Claude freezes on a random token
<img width="609" height="56" alt="Image" src="https://github.com/user-attachments/assets/40d756b7-d220-41b8-a0ef-56666f85fff0" />
<img width="1052" height="57" alt="Image" src="https://github.com/user-attachments/assets/69c7ee68-50de-4d54-9fa9-0a07e505a1ba" />
Is this a regression?
Yes, this worked in a previous version
Last Working Version
_No response_
Claude Code Version
2.1.119 (native, win32-x64)
Platform
Anthropic API
Operating System
Windows
Terminal/Shell
PowerShell
5 Comments
Found 3 possible duplicate issues:
This issue will be automatically closed as a duplicate in 3 days.
🤖 Generated with Claude Code
Closing for now — inactive for too long. Please open a new issue if this is still relevant.
This is still happening at a massive scale - I can't go through a single session without facing this dozens of times. How come such an high-priority bug is not planned ? Every related ticket is also closed as not planned, this is absolutely mad.
Can a human being at Anthropic step up to make sure this gets properly addressed ?
Hmm. In my case, this issue no longer occurs @Overdrivr. I just get the impression that the models now take longer to think (while still consuming tokens). It also seems to me that the quality of the responses has deteriorated quite noticeably.
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.