That particular benchmark is running st on macOS using Apple's XQuartz emulation layer. That's naturally a lot slower than running st on the operating system it

That particular benchmark is running st on macOS using Apple's XQuartz emulation layer. That's naturally a lot slower than running st on the operating system it was written for (see the "linux-st" line on the chart). And if I'm reading the table in that post correctly, st still used one-tenth the memory of any other terminal tested. But yes, if you use macOS, st is probably not the terminal for you.