In general, we can say that bigger clusters will process big queries faster and improve throughput. You should keep in mind that bigger clusters will not help in serving more concurrent queries. Endpoints will auto-scale (add a new cluster) when processing approximately 10 parallel requests.