Komodo 13 is out now. A miniacule 10 ELO over version 12.
I won’t comment on the Monte Carlo version other than to say perhaps someday that will exceed regular chess engines.
According to CCRL engine rating website the best version of Komodo 11 beat out Komodo 12 by a scant 2 elo.
The bottom line is that without a new paradigm in creating chess engines, it looks like we’re hitting a ceiling.
That being said, I’ve long looked at the current chess engine rating sites with a large grain of salt due to how little time chess engines are allowed.
The never specify the time controls, but CCRL is suppose to mimic an Athlon X2 @ 15 minutes.
I won’t include Leela Chess Zero, due to it being a distributed computing project.
Perhaps in the future when it, or a newer more efficient program, could be be independently installed on a single computer, sans any online connectivity, it might be a viable chess engine.
One has to wonder with machine AI, if it would be possible to do that on a personal computer if a large chunk of the processing power could utilize high end graphic cards.
Lc0 does primarily use graphics cards for its optimal performance. Chess.com did run a computer chess championship running engines on a single computer. I think Lc0 did beat Stockfish in their last computer tournament