IsItNerfed?

IsItNerfed? is a continuous LLM evaluation platform that helps developers and AI users track whether large language models are performing better or worse over time. The platform combines community-driven "vibe checks" with automated metrics to provide real-time insights into AI model performance changes.

The platform addresses a common concern in the AI community: whether LLM providers silently degrade or "nerf" their models. By aggregating user feedback and running standardized coding tasks, IsItNerfed? provides transparency into model performance fluctuations.

• Vibe Check System allows users to vote on whether specific AI agents feel "Smarter," "Same," or "Nerfed" compared to previous experiences, with real-time aggregation of community sentiment over 24-hour periods.

• AI Agent Tracking monitors popular coding assistants including Claude Code, Codex CLI, and Gemini CLI, displaying hourly and daily performance indicators based on user votes.

• Metrics Check continuously runs standardized coding tasks against LLMs to objectively measure failure rates over time, with lower scores indicating better performance.

• Historical Charts powered by TradingView display failure rate trends over 7-day and 30-day periods for models like Claude Code (Sonnet 4.5), Claude Code (Sonnet 4), and GPT-4.1.

• Model-Specific Tracking provides separate performance metrics for different model versions, allowing users to compare how specific models perform on coding tasks.

To get started, simply visit the website and participate in vibe checks by voting on how AI agents are performing for you. View the metrics dashboard to see objective failure rate data and historical trends. The platform requires no account creation for basic usage and provides immediate access to community sentiment and performance data.