Hi there kristianandersen
Thanks for using QandA platform
looks like a transient issue with the Document Translation service rather than something wrong on your side. The pattern of a few 403s followed by a successful 200 a second later means the backend is doing a token/permission refresh or hitting a brief capacity throttle. Since you're seeing it across different tenants and resource groups, that lines up with a service-side hiccup. check that you're sending the right region header if you’re using key auth, and add retry logic with back-off since the service does occasionally return temporary 403s like this. I’d also keep an eye on Service Health for West Europe, but based on the behavior it should settle on its own.
If this helps kindly accept the answer thanks much,