[wozzp]

Not in practice. Computing the absolute minimum solution for every possible position is computationally infeasible for a web-based solver. This uses Kociemba’s two-phase algorithm instead, which produces very efficient solutions, usually close to optimal, without requiring enormous amounts of time and memory.

[dtjc]

Why infeasible? cube20.org says "a good desktop PC" can optimally solve 0.36 random positions per second. And that's from the year 2010. I don't see why a web-based solver 16 years later should be much slower.

And apparently vcube optimally solved six cubes per second eight years ago already: https://github.com/Voltara/vcube