Users are reporting that Anthropic's Claude Sonnet 4.6, particularly within its VS Code extension, may be experiencing a decline in efficacy. Initially, the model incorrectly stated that chat history was not persisted, but later corrected itself to acknowledge local storage within the `.claude/projects/` directory on Windows. This incident, coupled with other users' observations of increasingly incorrect answers over time, has led to speculation about a potential degradation in the model's performance. AI
IMPACT User experiences suggest potential issues with model consistency and memory, impacting its reliability for coding tasks.
RANK_REASON User reports and discussions about a model's perceived performance degradation.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →