That’s a trade off, now you need multiple model calls for every single request