AIStupidLevel

Independent AI benchmarking with continuous drift detection, find out when an AI model silently degrades. We track GPT-5, Claude Opus 5, Gemini 3, DeepSeek V4, Kimi and GLM with live performance scores, CUSUM drift alerts and historical charts. Plus a smart API router that sends every request to the best-performing model, with usage monitoring and budget controls behind one API key.