3 ms·

I’ve heard Microsoft customers got burned by this. GPU availability and inference workloads is prioritized for OpenAI