The thing is that the entry level of provisioned throughput is so high! I just want a reliable model experience for my small Dev team using models through Vertex but I don't think there's anything I can buy there to ensure it.