Dragon by Komodo Chess

From Chessprogramming wiki
Jump to: navigation, search

Home * Engines * Komodo * Dragon

Dragon by Komodo Chess [1]
Logo at TCEC

Dragon by Komodo Chess, (KomodoDragon, Komodo Dragon)
a version of Komodo which combines an NNUE based evaluation with Komodo's search, either classical alpha-beta as well as MCTS. Dragon was announced on November 02, 2020, and released as Dragon by Komodo Chess on November 10, 2020. Strength improvements for both search modes were given in the 150 (4 threads) to 190 (one thread) Elo range [2].

Dragon is the work of Larry Kaufman and Mark Lefler, who superseded original Komodo author Don Dailey after his death on November 22, 2013, further supported by Dmitry Pervov and Dietrich Kappe concerning the NNUE inference code and training data. The neural network was trained supervised by chess Grandmaster Larry Kaufman, along with reinforcement learning considering billions of positions and evaluations from Komodo games. Chess.com supported the project financially [3].

Dragon 3.3 is expected to be the final version of Dragon after Mark Lefler and Larry Kaufman retire from running Komodo Chess, turn it over to Chess.com[4], and remain as consultants through 2025.[5]

Tournament Play

Dragon by Komodo Chess, at TCEC called KomodoDragon, played the Premier Division of TCEC Season 20, and became strong third behind LCZero and Stockfish in 2020/2021. KomodoDragon became third again at TCEC Season 21, winning the Infrafinal, and became runner up behind Stockfish at TCEC Season 22 in 2022.


Forum Posts


Re: Dragon by Komodo Chess by Larry Kaufman, CCC, November 14, 2020
Re: Speculations about NNUE development (was New engine releases 2020) by Connor McMonigle, CCC, November 12, 2020 » NNUE, Seer, Halogen




External Links