Arch is an open-source gateway specifically designed to handle prompts, leveraging fast Large Language Models (LLMs) for processing prompts and seamlessly integrating with backend systems. Built on Envoy, it supports any application language and offers quick deployment and transparent upgrades. It includes a variety of features such as traffic management, frontend/edge gateways, monitoring, and end-to-end tracing to help developers create fast, robust, and personalized GenAI applications.