Anthropic addresses Claude performance issues and usage limit changes

Anthropic has faced scrutiny regarding the performance and usage policies of its AI model, Claude. In early 2026, the company addressed technical issues with Claude Code, an AI tool for programming. Users reported degraded performance, including less attentive code reading and repetitive steps. Anthropic released a technical report confirming that three overlapping modifications deployed between March and April affected reasoning levels and response lengths, though the company stated that model weights and the API remained unchanged.

Additionally, Anthropic has encountered backlash over changes to Claude Code usage limits. While the company announced a permanent 25% increase in weekly limits, the move follows the expiration of a 50% promotional bonus. For current paid users, this transition results in a net 17% reduction in capacity compared to their current promotional levels. Following criticism that its initial communication was misleading, Anthropic issued a clarification acknowledging the 17% reduction relative to current usage.