There are some benchmarks[1] in the molten repo as well as instructions for running them yourself. On my machine it ends up being faster than Flask and nearly as fast as Falcon.

[1]: https://github.com/Bogdanp/molten/tree/master/benchmarks