[BUG] Claude Code 2.1.119 silently switches you to claude-opus-4-7[1m] even if you pick claude-opus-4-7

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?

I'm on the newest version of claude code (2.1.119) and picking claude-opus-4-7 (200k context window) in the model selector silently routes you to the claude-opus-4-7[1m] variant. This just started happening today, and no, I haven't changed anything in my settings.

settings.json

<img width="1914" height="717" alt="Image" src="https://github.com/user-attachments/assets/c8c8b6b6-a4d6-47d9-ac09-fa04234bb926" />

What Should Happen?

claude --model claude-opus-4-7 should use the 200k model (claude-opus-4-7), not claude-opus-4-7[1m].

Error Messages/Logs

victor@victor:~/Documents/GanbaruAI$ claude --version
2.1.119 (Claude Code)
victor@victor:~/Documents/GanbaruAI$ echo '{"env":{"CLAUDE_CODE_EFFORT_LEVEL":"max"},"attribution":{"commit":"","pr":""},"permissions":{"defaultMode":"auto"},"enabledPlugins":{"rust-analyzer-lsp@claude-plugins-official":true},"showThinkingSummaries":true,"skipDangerousModePermissionPrompt":true,"preferredNotifChannel":"terminal_bell","skipAutoPermissionPrompt":true,"model":"claude-opus-4-7"}' > ~/.claude/settings.json
victor@victor:~/Documents/GanbaruAI$ claude --model claude-opus-4-7
 ▐▛███▜▌   Claude Code v2.1.119
▝▜█████▛▘  Opus 4.7 · Claude Max
  ▘▘ ▝▝    ~/Documents/GanbaruAI

❯ /context
  ⎿  Context Usage
     ⛁ ⛁ ⛁ ⛁ ⛁ ⛀ ⛀ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   Opus 4.7
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   claude-opus-4-7
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   26.1k/1m tokens (3%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   26.1k/1m tokens (3%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   Estimated usage by category
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ System prompt: 9.1k tokens (0.9%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ System tools: 11.1k tokens (1.1%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ Memory files: 5k tokens (0.5%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶   ⛁ Skills: 889 tokens (0.1%)
     ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛶ ⛝ ⛝ ⛝ ⛝ ⛝ ⛝ ⛝   ⛁ Messages: 13 tokens (0.0%)
                                               ⛶ Free space: 940.9k (94.1%)
                                               ⛝ Autocompact buffer: 33k tokens (3.3%)

Steps to Reproduce

1. echo '{"env":{"CLAUDE_CODE_EFFORT_LEVEL":"max"},"attribution":{"commit":"","pr":""},"permissions":{"defaultMode":"auto"},"enabledPlugins":{"rust-analyzer-lsp@claude-plugins-official":true},"showThinkingSummaries":true,"skipDangerousModePermissionPrompt":true,"preferredNotifChannel":"terminal_bell","skipAutoPermissionPrompt":true,"model":"claude-opus-4-7"}' > ~/.claude/settings.json
2. claude --model claude-opus-4-7
3. /context

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

2.1.118

Claude Code Version

2.1.119

Platform

Anthropic API

Operating System

Ubuntu/Debian Linux

Terminal/Shell

Other

Additional Information

settings.json

If you want to force the actual 200k model, add this to ~/.claude/settings.json:

"env": {
  "CLAUDE_CODE_DISABLE_1M_CONTEXT": "1"
}

P.D.: Even when I keep usage under 200k tokens on the 1M variant, it feels dumber than the 200k model. And after their April 23 postmortem admitting to three separate quality regressions in Claude Code, call me paranoid, I personally feel the 1M variant is quantized or something in the Claude Code harness behaves differently with it, That's why I'm not just setting CLAUDE_CODE_AUTO_COMPACT_WINDOW.