Daily AI model reliability tracker for workload routing

Someone should make an app where AI X evaluations are actually proven by test and not some vague reposting or claims. I would personally pay around 2$ to read and use it per month.

I'd happily pay for one app that automatically picks the right model for each task.

Has anyone built a system that tracks how smart AI models are from day to day?

I would pay for a tool that told me which days I should use which models instead of realizing an hour in that the model I'm using is an absolute retard today