Stockfish is a free, open-source chess engine trusted by grandmasters and chess platforms worldwide, built around NNUE (Efficiently Updatable Neural Network) evaluation since 2020 – a neural network trained on millions of positions at moderate search depth, engineered to run efficiently on ordinary CPU hardware rather than requiring specialized AI accelerators.
Efficient CPU evaluation is the technically significant part – most neural network applications assume GPU acceleration, but NNUE exploits the fact that only parts of the network need updating after a typical chess move, using low-precision integer arithmetic optimized for standard processors. That engineering choice is why Stockfish’s neural evaluation could run inside the same fast alpha-beta search the engine already used, rather than requiring a complete architectural rebuild – and why the hybrid decisively won the Top Chess Engine Championship (TCEC) once adopted, firmly establishing classical search plus neural evaluation as the strongest engine paradigm.
Stockfish is released under GPLv3, with source code publicly available on GitHub and development driven by an open community rather than a commercial vendor. It integrates into chess GUIs, analysis tools, and research applications across desktop, web, and mobile platforms, completely free with no signup required.
