All of our results are collected in a GitHub runner, which is good for consistency. It would be potentially interesting to let the user run benchmarks on their own machine for comparison, to see how it compares.