Leela Chess Zero
Open-source neural network chess engine trained through self-play, offering a distinctive style compared to classical engines.
What is Leela Chess Zero?
Leela Chess Zero (Lc0) is an open-source chess engine that uses deep neural networks and Monte Carlo tree search, inspired by DeepMind's AlphaZero. Unlike traditional engines such as Stockfish that rely on handcrafted evaluation functions, Lc0 learned entirely through self-play, starting from random moves and developing its own understanding of chess strategy.
This approach gives Lc0 a distinctive playing style: it tends to find long-term positional sacrifices and unconventional plans that classical engines may miss. It is one of the top-rated engines in independent computer chess rating lists and has been a consistent contender in TCEC supertournaments alongside Stockfish.
Lc0 is free, open source (Apache 2.0 license), and runs on GPU or CPU hardware. It integrates with UCI-compatible GUIs. Running it efficiently requires a reasonably modern GPU; for players without dedicated GPU hardware, the web-based analysis options from Lichess or Chess.com (which use Stockfish) may be more practical.