Anthropic’s cutting-edge AI tool, Claude Code, has revolutionized the coding landscape. Utilizing Anthropic’s advanced Claude models, this AI tool can generate hundreds, and in some cases, thousands of lines of code from a single prompt. However, recent user feedback suggests a decline in performance, with users expressing concerns about Claude Code’s diminishing capabilities.
On various platforms, users have voiced frustrations, questioning the functionality of Claude Code. Comments range from “Is Claude broken?” to “Claude code is no longer usable.” Some even likened the AI to a negligent employee, illustrating scenarios where it acts independently, deviates from instructions, and shifts focus arbitrarily.
This isn’t the first time Anthropic has faced scrutiny. Previously, the company faced backlash for attempting to restrict access to Claude Code under its $20 Pro plan.
In response to user complaints, Anthropic, led by Dario Amodei, addressed the issues promptly. The company emphasized its commitment to maintaining model quality and released a patch on April 20 to rectify three identified issues. By resetting user limits, Anthropic aimed to enhance accessibility to Claude Code.
Boris Cherny, head of Claude Code, acknowledged the gravity of the situation and emphasized the thorough investigation undertaken to identify and resolve underlying issues. The company clarified that the recent problems impacted Claude Code, the Claude Agent SDK, and Claude Cowork, while the API and inference layer remained unaffected.
The root of the problem, as highlighted by Anthropic, stemmed from three key changes made to Claude. Firstly, a shift in default reasoning settings from “high” to “medium” on March 4 aimed to expedite AI responses but was met with user dissatisfaction. Further alterations on March 26 led to the inadvertent deletion of older session data, causing Claude to appear forgetful and repetitive. Lastly, modifications to the AI’s verbosity on April 16, combined with other prompt adjustments, resulted in a decline in coding quality.
Anthropic assured users that all issues had been addressed comprehensively, ensuring optimal performance from Claude Code moving forward.