Hey guys, I work on Hyperlane and wrote this article.

The benchmark is open source for anyone to try and it measures whatever editors are on the machine rather than being wired to ours, with a fresh profile every launch.

The speed difference is quite small, but we are hoping to make further cuts and optimizations to get it closer to native speeds in the future.