"No token was detected" when using Hosted inference API

Seems we all have the same issue.
Has someone found a solution for it?
Who could we contact about this?