I believe everyone should run their own evals on their own tasks or use cases.

Shameless plug, but I made a simple app for anyone to create their own evals locally:

https://eval.16x.engineer/