[BUG] Unexpected use of Claude Sonnet 4

Environment

• Platform (select one):
• [ ] Anthropic API
• [X] AWS Bedrock
• [ ] Google Vertex AI
• [ ] Other: <!-- specify -->
• Claude CLI version: 1.0.69
• Operating System: macOS 15.5
• Terminal: <!-- e.g. iTerm2, Terminal App -->

Bug Description

Despite being configured with Claude 3.7 Sonnet, claude ends up using Claude Sonnet 4.0

Steps to Reproduce

1. Start claude
2. Ask what model it is using
3. Start a sub-agent and have it tell you what model it is using
4. Run /cost

Expected Behavior

In both cases the model should be reported as Claude 3.7 Sonnet and /cost should reflect that.

Actual Behavior

First Claude 3.7 Sonnet is reported as the model.
Then Claude Sonnet 4.0 is reported.
The costs show usage of both models (4.0 usage is improperly shown as "claude-sonnet" with no version. Possibly tracked in issue #4294)

Additional Context

<!-- Add any other context about the problem here, such as screenshots, logs, etc. -->

<img width="767" height="429" alt="Image" src="https://github.com/user-attachments/assets/d88ba1b1-b2ce-499e-a48f-20deca35e1fc" />