Does anyone know how to mitigate this?

That usually points to rate limits or temporary API overload. You can mitigate it by adding retries with exponential backoff, request queuing, and fallback handling so the experience doesn’t break for users. I can help you set up a more resilient flow if you share your current stack. @MarcPP

Yeah but this happened multiple times, i can not implement exponential backoff or request queuing since this issue is on VAPI's side, and fallback handling is already kind of implemented by providing fallback models but it doesnt switch to a fallback model instead it cancels the call