Chess Engines Diary

FunComp  a UCI chess engine written in C++23 from the ground up: a classical evaluation (PeSTO's tables) and a modern alpha-beta search in which every feature was proven in games before it was kept. Version 1.0 is estimated at about 3030 on the CCRL Blitz scale (3031 ± 14 from a gauntlet of 2,040 games against 17 rated engines from eight families; an estimate on our hardware, not an official CCRL rating). Single-threaded. Author: James Coit FunComp 1.0 download   mirror Claudy  a UCI chess engine written from scratch in C++20, with an NNUE evaluation trained only on its own self-play games. Author: MS Kim Claudy 4.1 download     mirror Fructosita is an open-source UCI chess engine written entirely in Rust. The project began on July 2, 2026 as a personal initiative to learn chess engine development from scratch. The name Fructosita was inspired by the author's undergraduate thesis on fructose, eventually becoming the engine's permanent identity. The long-te...

Today marks the start of the CEDR Chess Engine League—Season 10/2026. We’re kicking off with League 1, and the tournament lineup is displayed in the graphic. The format is a round-robin, with each pair of engines playing four games. All engines have their CEDR ratings set as of October 1, 2026. More information about the league can be found on the LEAGUE page.

  Casanchess -  UCI compatible chess engine written in C++. The NNUE evaluation is trained on millions of positions from auto-games and randomly generated FENs, evaluated with its own classical evaluation upto depth 7. Author: Carlos Sanchez Mayordomo  Rating CEDR=3327 CasanChess what's new? A much stronger NNUE. I spent the last month (++) improving the whole process: data generation, binary format, trainer, architecture... resulting in +230 self-Elo gain only from the new NNUE. Moreover, now the evals are normalized (+1.0 eval corresponds to 75% expected score), and the endgame knowledge improves since the game results are used in the training, not only the evaluations. Trained from 170M positions from self-played games at 250k nodes. More info: #109 New UCI option: Ambition. The NNUE also estimates the drawishness of the position. Casanchess uses this information to modify the search score and play less drawish continuations. The default value 15 reduced draws from 48...

  What is Ferz? The built-in Ferz engine is designed to play in an aggressive, human-like style at approximately 2250 ELO. It doesn't just find the strongest move — it fights for the initiative, launches kingside attacks, sacrifices material for activity, and creates the kind of tactical complications that make chess exciting. Supports classical chess, computer tournaments, game analysis, opening books, and more — all in one self-contained program with a clean dark-themed interface. More information about the program and files for download can be found here – HERE .

Luna Chess Engine Luna is a high-performance chess engine written in Rust, designed for efficiency and strength. It combines classical bitboard-based move generation with a state-of-the-art NNUE (Efficiently Updatable Neural Network) evaluation, allowing for deep positional understand in. Author: Spunc595 Luna Chess-Engine (LCE) JA what's new? A network trained by the author on public data This is the first release whose embedded network (resources/net.bin) was trained by the author, with his own training pipeline, on public Leela Chess Zero data. It was trained with bullet ,on 1 billion positions (about 914 million unique by exact board and side-to-move match; 8 epochs) in the king-bucketed 768x4 mirrored architecture ported from akimbo: 1024 hidden neurons, squared clipped ReLU. The file format and size are unchanged. The architecture and the AVX2 inference kernel are still akimbo's. Only the trained weights changed. Training data Leela Chess Zero self-play games (runs T77 an...

KnightX - UCI/winboard chess engine Author- Christophe Jolly  Rating Chess Engines Diary CEDR=2856 KnightX ,  a Chess Engine Communication Protocol compliant chess engine by Christophe Jolly, written in C, and alias and successor of the native Windows program Techno Chess. KnightX, supporting WinBoard, was first published in December 1999, and since 2004, KnightX 1.86 runs under Linux via XBoard. KnightX played the Massy 2003 French Programmers Tournament, and the FCCC 2003 French Computer Chess Championship. After a long break from 2005 with KnightX 1.92, Christophe Jolly had his comeback in February 2019, when he released Knightx 1.93, continuing with subsequent versions. ( source: Chess Programming Wiki). What has changed in the new version? Neural network evaluation using the Cerebrum library of David Carteau. The executable uses the AVX2 instructions set (Intel processors since 2013, AMD processors since 2015). The archive also contains an executable that doesn't use ...

  JustBot is my first attempt at creating a UCI chess engine with magic bitboards without any agentic or LLM assisted coding. Author: Hassan Fakih  Rating Chess Engines Diary CEDR=3721 JustBot what's new? Results Against Previous Version STC: Elo   | 104.58 +- 6.53 (95%) Conf  | 8.0+0.08s Threads=1 Hash=32MB Games | N: 3008 W: 993 L: 114 D: 1901 Penta | [0, 50, 636, 707, 111] https://ob.justchess.xyz/test/823/ LTC: Elo   | 81.65 +- 9.66 (95%) Conf  | 60.0+0.60s Threads=1 Hash=128MB Games | N: 1014 W: 255 L: 21 D: 738 Penta | [0, 10, 271, 208, 18] https://ob.justchess.xyz/test/824/ 4T STC: Elo   | 99.03 +- 10.37 (95%) Conf  | 8.0+0.08s Threads=4 Hash=128MB Games | N: 1016 W: 299 L: 17 D: 700 Penta | [0, 9, 236, 235, 28] https://ob.justchess.xyz/test/825/ 4T LTC: Elo   | 74.43 +- 8.91 (95%) Conf  | 60.0+0.60s Threads=4 Hash=512MB Games | N: 1000 W: 220 L: 9 D: 771 Penta | [0, 3, 296, 188, 13] https://ob.justchess.xyz/t...

  Rusty Rival -  UCI (Universal Chess Interface) chess engine written in Rust. Author:Chris Moreton Rating Chess Engines Diary CEDR=3235 The engine can be about 50% faster if compiled on the machine on which it will be run in order to take advantage of cpu-specific instructions.  Rusty-rival what's new? Two stacked changes from the September 2026 search-efficiency audit (Notion: Search Efficiency Audit — September 2026; Linear NET-1154), merged as PR #86. Bug fix NET-1187 — the singular-extension verification search no longer searches the move it excludes. The excluded TT move was filtered from the generated move lists but not from the hash-move path, and the TT probe inside the verification returns the same entry, so every verification searched the excluded move first (98 of 98 at depth 18 on the reference position). The extension had therefore never fired. The hash-move path is now guarded, and a verification that leaves no legal alternative reports a fail-low instead o...