I've been writing JSON-REST APIs for a number of years, all the time wanting to have tooling that validates the JSON sent and received and automatically generates documentation. I finally have this with Quart-Schema, and this article explains how I got there.
Type hints in Python 3 are executed at import time at a computational cost. This cost can removed in Python >= 3.7 via from __future__ import annotations, see PEP-563. The startup time when importing quart is measured both with and without the future import with no difference observed.
I've proposed a change to Flask to support async(io) view functions following on from a recent change to Quart to support (non-blocking) sync view functions. This article explains this approach and why Flask and Quart are necessarily complimentrary projets.