[BUG] Claude-code wronlgy sending each user request twice to Anthropic model

tcpdump.txt

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?

Claude-code request the sever twice for each user prompt.

First, it makes a call for claude-haiku to get the quota/auth. This is ok.

Then, for each request, it calls:

1) claude-opus with max_tokens 32000, stream=true
2) claude-opus with max_tokens 21333, stream=false
I don't know if the two copies are billed or only one, probably both.

Related with https://github.com/anthropics/claude-code/issues/21969

What Should Happen?

Claude-code should request the server only once per user prompt.

Error Messages/Logs

Steps to Reproduce

1. Run claude
2. Say "hi"

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.63

Platform

Anthropic API

Operating System

Ubuntu/Debian Linux

Terminal/Shell

Other

Additional Information

tcpdump.log