Out of the box, Phoenix applications respond to simple http requests in times measured in microseconds. What appreciable improvements from that do you get with Python? And considering how much of your total request time is not processing by the language (db calls, network latency, etc.), why would you decide on a language purely on the minor speed improvement of a small part of the overall picture? I’ll gladly trade what might amount to a few ms of request time for the concurrency model, scalability, and latency characteristics of Elixir and the BEAM.

To each their own, I suppose.

Yeah I don't get this argument either. It's like people are still stuck in micro-benchmark-land. Sub-1ms responses are the norm in Phoenix and I tend to never get anything above 100µs for LiveView messages, sure, they're noops, but calling it slow is .. well, strange.

Sure, you don't want to do number crunching in pure Elixir, but I'm always curious as to the actual needs people have, once being one of the "performance is everything" kind of developers myself.

For number-crunching there's Nx. As far as I can tell it's a solved problem, with easy handling of things like batching over clusters of GPU:s and things like that as a bonus.