Zephr aims to deliver dynamic HTML in under 100ms and supports up to 15,000 TPS per region with low latency.

In countries where Zephr has a proxy cluster running, we aim to serve dynamic HTML in less than 100ms – although there are many factors involved in site performance, some of which are outside of our control. For raw decision API requests, Zephr has been tested up to 15,000 TPS per region with an average latency of 9ms and a p99 of 16ms.