Comment by HawtAds

It's very much real but not necessarily malicious. We track upstream providers pretty closely. Sometimes it's a just matter of a single GPU runtime layer bug/update to break inference outputs. The model weights don't necessarily change/get quantized.