Check out these benchmarks. https://github.com/rosetta-rs/template-benchmarks-rs

Particularly the implementation of the benchmark for each crate.

When you say “best in class performance”, do you have particular benchmarks in mind?

I only found this one: https://github.com/rosetta-rs/template-benchmarks-rs

Askama was neither fastest not slowest. Purely based on benchmarks sailfish seems to be interesting as well