[BUG] Claude Cowork defaults to 1M context on Pro plan, throwing "Usage credits required" error (Ignores env variables)
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?
The Claude Cowork desktop app is automatically defaulting to the 1M token context window, even on fresh sessions. Since I am on the standard Pro subscription and do not have additional pay-as-you-go Usage Credits enabled, the app immediately blocks my prompts with an API error. The biggest issue is that the app completely ignores the environment variable CLAUDE_CODE_DISABLE_1M_CONTEXT=1 in ~/.zshrc and the config.json file.
What Should Happen?
The application should use the standard context limit included in the Pro plan without forcing the 1M context. Furthermore, it should respect local configuration files and environment variables (like config.json or .zshrc) to disable the 1M context when set by the user.
Error Messages/Logs
API Error: Usage credits required for 1M context · turn on usage credits at claude.ai/settings/usage, or use --model to switch to standard context
Steps to Reproduce
Open the Claude Cowork desktop app on macOS.
Start a new chat or continue a standard task.
Send any prompt.
The system attempts to route to the 1M context model and throws the API error immediately.
Attempting to bypass this by adding export CLAUDE_CODE_DISABLE_1M_CONTEXT=1 to the Terminal (~/.zshrc) or to the local config.json fails, as the app ignores these settings.
Claude Model
Sonnet (default)
Is this a regression?
Yes, this worked in a previous version
Last Working Version
2.1.149
Claude Code Version
2.1.161
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
<img width="1728" height="1117" alt="Image" src="https://github.com/user-attachments/assets/c2cf5dc1-2170-45b0-bbc1-75c955172926" />
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗