[BUG] Claude Code via Vertex AI is unbearably slow (4 minute delay before responding)

Preflight Checklist

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

What's Wrong?

When using Claude Code on Vertex AI, responses are very slow.

Since yesterday, I have to wait 3-4 minutes before the model responds each turn. Sometimes more. In this screenshot I wait 8 minutes and still no response.

<img width="841" height="94" alt="Image" src="https://github.com/user-attachments/assets/88f7cbf6-74d0-4be8-a9f3-fbb2888ccece" />

[!NOTE] - This does not happen when using the Claude Pro subscription, so it seems to be a connectivity problem between Vertex AI and Anthropic. - I tried switching from global region to different regions, to no avail.

What Should Happen?

Model should respond fast.

Error Messages/Logs

Steps to Reproduce

Use Claude Code with

export CLAUDE_CODE_USE_VERTEX=1

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

v2.0.37

Platform

Google Vertex AI

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_