AIStupidLevel - Live AI benchmarks, drift alerts, and smart model routing

Hey Product Hunt! I’m Ionut, the founder of AIStupidLevel. AI models don’t remain consistent. The same model and prompt can produce materially different results from one day to the next, yet users rarely receive any warning when performance changes. I built AIStupidLevel to make those changes measurable. It continuously evaluates leading AI models across coding, deep reasoning, tool use, cost, and latency. It detects performance drift and powers an OpenAI-compatible Smart Router that uses current benchmark data to choose the right model for each request. So far, AIStupidLevel has: Run 167,000+ benchmarks Processed 88M+ tokens Measured more than 80 models Reached 120,000+ active users without paid marketing You can explore live and historical model performance, identify degradation, test models using your own API keys, and route production workloads through a single compatible API. My goal is to build an independent intelligence layer for AI - one that makes model quality transparent, continuously monitored, and actionable. I’d love to hear what you think, which models you want us to monitor next, and what you would need from the Smart Router to use it in production.