Anthropic users across online forums are raising complaints that the Claude AI model has suddenly declined in performance. Over recent weeks, users on X, GitHub, and Reddit have been sharing anecdotes, benchmarks, and prompts to identify what changed. One AMD senior director wrote that Claude "has regressed to the point it cannot be trusted to perform complex engineering."
Speculation centers on whether the model was deliberately scaled back—what users call "nerfed"—to control costs or redirect scarce computing power toward Anthropic's more advanced Mythos model. The backlash arrives as the company tests this more powerful system, raising questions about whether cutting-edge AI is becoming less accessible even as it grows more capable.