Anthropic: Fixes Claude Code Performance Degradation, Resets Usage Limits for All Subscribers

Odaily reported that according to official sources, Anthropic's previous feedback regarding a decline in Claude Code's quality stemmed from three independent changes involving Claude Code, the Claude Agent SDK, and Claude Cowork. Specific issues included: On March 4, the default reasoning intensity was reduced from high to medium, leading to a drop in intelligence level; this setting was restored on April 7. A bug in the cache optimization feature launched on March 26 caused old reasoning records to be incorrectly cleared, resulting in the model exhibiting forgetfulness and repetition; it was fixed on April 10. System prompts added on April 16 to reduce redundancy harmed programming quality and were revoked on April 20. Currently, all affected models have returned to normal, and Anthropic has reset the usage limits for all subscribers.