We use flask backed by apache mod_wsgi running in containers or VMs at work for our micro Services and it's quite performant. I've run stress tests with this setup using freeBSDs siege command (awesome btw) and 1000+ concurrency and it ran without a hiccup.
I can't say at what level it would break, but I'm confident it would perform as well as django (and it's designed to be thread safe so assuming your following the 12factor app principles And use the process model/attached storage for persistence running elsewhere, you can run multiple threads, apache vhost multi threading notwithstanding.
I used Nginx as a HTTP proxy and gunicorn for WSGI. It worked decently well, although it was pretty resources intensive (CPU, RAM).
Today you can use something like Caddy[0] and Gunicorn[1]
[0] https://caddyserver.com
[1] http://gunicorn.org