Users of Claude Code, specifically with the Claude 4.6 Opus model, have reported instances where the application appears to stop "thinking" during conversations. Investigation revealed that the underlying model is still performing its reasoning processes, but a recent update to the desktop application has caused it to omit the `--thinking-display` flag. This change results in the application discarding the generated thinking text, making it seem as though the model is no longer engaged in deep thought, even though the reasoning is still occurring internally. AI
IMPACT A bug in the Claude Code desktop application may cause users to perceive a decline in Claude 4.6 Opus's reasoning capabilities, despite the model functioning correctly.
RANK_REASON User-reported issue with a specific application version of a model.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →