| AI |
Version |
Author |
Download link |
Comment |
| RAPFI |
2025 (0.43.01) |
Haobin Duan, Zhiyang Hang, and other contributors listed in https://github.com/dhbloo/rapfi/blob/master/AUTHORS |
Download |
New NNUE architecture featuring a reduced weight size and lower memory footprint. Added support for AVX-VNNI, AVX-512, and AVX-512 VNNI instruction sets. |
| RAPFI |
2024 (0.40.03) |
Haobin Duan, Zhiyang Hang, Rapfi Contributors, Stockfish Developers, and Pikafish Team |
Download |
NNUE architecture has been updated and the engine has been tuned. Bug fixes. |
| RAPFI |
2023 (0.39.01) |
Haobin Duan, Zhiyang Hang, Rapfi Contributors, Stockfish Developers, and Pikafish Team |
Download |
The NNUE has been updated with newer architecture and trained weights, and the engine has been tuned with SPSA. General search has been improved, and database functionality has been introduced that allows recording database files which is compatible with Yixin's format. |
| RAPFI |
2022 (0.34.05) |
Haobin Duan, Zhiyang Hang |
Download |
Rapfi is now equipped with an NNUE (Efficiently updatable neural network) trained on games played by Katagomo. Although NNUE runs with a lower speed, its accuracy on evaluation brings notable strength improvement. Classical evaluation is improved via automatic tuning, and combines with NNUE in a hybrid way. Search and speed is also improved. This version is for the unlimited league. |
| RAPFI |
2022 (0.34.05) |
Haobin Duan, Zhiyang Hang |
Download |
Rapfi is now equipped with an NNUE (Efficiently updatable neural network) trained on games played by Katagomo. Although NNUE runs with a lower speed, its accuracy on evaluation brings notable strength improvement. Classical evaluation is improved via automatic tuning, and combines with NNUE in a hybrid way. Search and speed is also improved. |
| RAPFI |
2021 (0.29.02) |
Haobin Duan |
Download |
Support for standard and renju rule is added; serveral bugs are fixed and some modification is made on evaluation; basic support for Yixin-Board is added. |
| RAPFI |
2019 (0.26.11) |
Haobin Duan |
Download |
Algorithm has been completely rewritten, and speed was improved. Some modifications were made on evaluation. |
| RAPFI |
2018 (2018.02) |
Haobin Duan |
Download |
|
| KATAGOMO |
2026 (2026_nano) |
Zhiyang Hang and contributors of Katago |
Download |
The 2026 version is approximately 200 Elo stronger than the 2021 version across all rule sets. In addition, the Caro rule is now supported. |
| KATAGOMO |
2021 (20210502) |
Zhiyang Hang and contributors of Katago |
Download |
Rewritten based on a new Katago version, so it has a CPU-only version now and will participate in all the classical leagues as well; added a strict VCF solver; new nets with VCF feature input. This version is used in the standard group. |
| KATAGOMO |
2021 (20210502) |
Zhiyang Hang and contributors of Katago |
Download |
Rewritten based on a new Katago version, so it has a CPU-only version now and will participate in all the classical leagues as well; added a strict VCF solver; new nets with VCF feature input. This version is used in the renju group. |
| KATAGOMO |
2021 (20210502) |
Zhiyang Hang and contributors of Katago |
Download |
Rewritten based on a new Katago version, so it has a CPU-only version now and will participate in all the classical leagues as well; added a strict VCF solver; new nets with VCF feature input. This version is used in the freestyle group. |
| KATAGOMO |
2021 (20210502) |
Zhiyang Hang and contributors of Katago |
Download |
Rewritten based on a new Katago version, so it has a CPU-only version now and will participate in all the classical leagues as well; added a strict VCF solver; new nets with VCF feature input. This version is used in the experimental tournament. |
| KATAGOMO |
2020 (1.3.1) |
Zhiyang Hang and contributors of Katago |
Download |
Katagomo is derived from Katago, a very efficient Go program with alphazero algorithm. The Gomoku version is modified and trained by Zhiyang Hang. KataGomo can reach very high level with several days of training with a few GPUs, maybe much stronger than Embryo 2019 with same time and same price computer. |
| ALPHAGOMOKU(MK) |
2026 (5.10.0) |
Maciej Kozarzewski |
Download |
Bugfixes and tuning. |
| ALPHAGOMOKU(MK) |
2025 (5.9.2) |
Maciej Kozarzewski |
Download |
New neural network architecture. |
| ALPHAGOMOKU(MK) |
2024 (5.8.0) |
Maciej Kozarzewski |
Download |
Speedup from using AVX512. Added OpenCL backend. Working support for YixinBoard. Small improvements in the alpha-beta search. Also added freestyle network for 15x15 board. |
| ALPHAGOMOKU(MK) |
2024 (5.8.0) |
Maciej Kozarzewski |
Download |
Speedup from using AVX512. Added OpenCL backend. Working support for YixinBoard. Small improvements in the alpha-beta search. Also added freestyle network for 15x15 board. This version is used in the unlimited tournament. |
| ALPHAGOMOKU(MK) |
2023 (5.6.0) |
Maciej Kozarzewski |
Download |
Added support for renju and caro. New networks trained on 5 million games. VCF solver was upgraded to alpha-beta serach. Added basic support for Yixinboard protocol. Also some speedups. |
| ALPHAGOMOKU(MK) |
2022 (5.3) |
Maciej Kozarzewski |
Download |
Now using Monte Carlo Graph Search. Fixes and optimizations in the VCF solver. Improved time management. Added several utilities like benchmarking or automatic configuration. |
| ALPHAGOMOKU(MK) |
2021 (5.0) |
Maciej Kozarzewski |
Download |
New training run of about 1 million games; included VCF solver in the search algorithm. This version is designed for the experimental tournament. |
| ALPHAGOMOKU(MK) |
2021 (5.0) |
Maciej Kozarzewski |
Download |
New training run of about 1 million games; included VCF solver in the search algorithm. This version is used in the classical leagues. |
| ALPHAGOMOKU(MK) |
2020 (4.0) |
Maciej Kozarzewski |
Download |
New implementation. It now uses some Gomoku-specific features as input to neural network. New training run of approximately 600k self-play games. |
| ALPHAGOMOKU(MK) |
2019 (3.0) |
Maciej Kozarzewski |
Download |
Completely new implementation. Network was trained for much longer than the previous one. Also, now it supports standard rule on 15x16 board. This version supports both freestyle and standard rules. |
| ALPHAGOMOKU(MK) |
2019 (3.0) |
Maciej Kozarzewski |
Download |
Completely new implementation. Network was trained for much longer than the previous one. Also, now it supports standard rule on 15x16 board. This version is used in the standard group. |
| ALPHAGOMOKU(MK) |
2019 (3.0) |
Maciej Kozarzewski |
Download |
Completely new implementation. Network was trained for much longer than the previous one. Also, now it supports standard rule on 15x15 board. This version is used in the freestyle group. |
| ALPHAGOMOKU(MK) |
2018 (1.0) |
Maciej Kozarzewski |
Download |
AlphaGomoku is based on AlphaGo Zero - MCTS with deep neural network trained from self play. |
| JAX |
2025 |
Kailong Jiang |
Download |
The Monte Carlo tree search plus neural network model solution from last year is used, but the models are more powerful and there are fewer of them, and the four rules (Freestyle, Standard, Renju, Standard Caro) with a board size of 15 use the same model; one engine supports all groups of Gomocup2025 competitions; and a memory problem in multi-threaded search is fixed. |
| JAX |
2024 |
Kailong Jiang |
Download |
Multiple rules (Freestyle, Standard, Renju, and Standard Caro), multiple board sizes, and multiple devices (CPU, CUDA, TensorRT) are supported; the MCTS and the self-play improvement loop have been optimized, and several methods from KataGo have been adopted; random balanced openings and historical Gomocup openings have been added to the self-play; the spatial position feature of patterns (e.g., Double Four) is fed into the neural network to improve the precision of policy and value predictions; depending on the rules, the neural network is trained with between 2 and 10 million games. |
| JAX |
2024 |
Kailong Jiang |
Download |
Multiple rules (Freestyle, Standard, Renju, and Standard Caro), multiple board sizes, and multiple devices (CPU, CUDA, TensorRT) are supported; the MCTS and the self-play improvement loop have been optimized, and several methods from KataGo have been adopted; random balanced openings and historical Gomocup openings have been added to the self-play; the spatial position feature of patterns (e.g., Double Four) is fed into the neural network to improve the precision of policy and value predictions; depending on the rules, the neural network is trained with between 2 and 10 million games. This version is used in the unlimited tournament. |
| JAX |
2023 |
Kailong Jiang |
Download |
Improved MCTS and CNN, self-play training from zero. |
| BARBAKAN |
2021 (1.0) |
Janos Wolosz and contributors of Katago |
Download |
To determine the best moves Barbakan is combining two kinds of searches: a relatively slow Monte Carlo search with NN evaluation and a fast surewin search on the search leaves. The Monte Carlo search and the NN training pipeline of Barbakan is based on go program KataGo. Barbakan supports freestyle rule. |
| EMBRYO |
2026 (1.2.3-9f111a5f) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Refactored quiescence search, better move ordering |
| EMBRYO |
2025 (1.2.2-30b506d1) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Better support for AVX2 instruction set. Some speedup in search. |
| EMBRYO |
2024 (1.2.1-2ee4b2f8) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Supports AVX2 instruction set. Fixed bug in move generator |
| EMBRYO |
2023 (1.2.0-2084105e) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Improvement in search pruning, fixed bugs in Caro5 variant. |
| EMBRYO |
2022 (1.1.1-381a3618) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
The evaluation also takes into account 3 + 2 and 2 + 2 patterns. C++ code ready to run on Android, release with GUI coming soon. |
| EMBRYO |
2021 (1.0.5-785859aa) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Fixed wrong forbidden points classification in Renju; fixed swap logic; some speedup. This version is used in the standard group. |
| EMBRYO |
2021 (1.0.6-66c9cd65) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Fixed wrong forbidden points classification in Renju; fixed swap logic; some speedup. This version is used in the renju group. |
| EMBRYO |
2021 (1.0.5-785859aa) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Fixed wrong forbidden points classification in Renju; fixed swap logic; some speedup. This version is used in the freestyle group. |
| EMBRYO |
2021 (1.0.5-785859aa) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Fixed wrong forbidden points classification in Renju; fixed swap logic; some speedup. This version is used in the experimental tournament. |
| EMBRYO |
2020 (1.0.5-7e4e7a1d) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Fixed wrong forbidden points classification in Renju, support for database format in Yixin format, pondering, new Caro version (like freestyle, except OXXXXXO is not a win), basic preparation for autotuning. This version is used in the freestyle group. |
| EMBRYO |
2020 (1.0.5-7e4e7a1d) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Fixed wrong forbidden points classification in Renju, support for database format in Yixin format, pondering, new Caro version (like freestyle, except OXXXXXO is not a win), basic preparation for autotuning. This version is used in the standard group. |
| EMBRYO |
2020 (1.0.5-7e4e7a1d) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
Fixed wrong forbidden points classification in Renju, support for database format in Yixin format, pondering, new Caro version (like freestyle, except OXXXXXO is not a win), basic preparation for autotuning. This version is used in the freestyle group. |
| EMBRYO |
2019 (0.6.4.2600) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
New support for Standard and Renju rules, new time management, fixed pattern matching and positional evaluation, fixed alpha-beta tree pruning, time-critical parts were rewritten using parallel SIMD instructions, overall speedup about 25%, some commands from Yixin GUI implemented. This version is used in the renju group. |
| EMBRYO |
2019 (0.6.4.2600) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
New support for Standard and Renju rules, new time management, fixed pattern matching and positional evaluation, fixed alpha-beta tree pruning, time-critical parts were rewritten using parallel SIMD instructions, overall speedup about 25%, some commands from Yixin GUI implemented. This version is used in the standard group. |
| EMBRYO |
2019 (0.6.4.2600) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
New support for Standard and Renju rules, new time management, fixed pattern matching and positional evaluation, fixed alpha-beta tree pruning, time-critical parts were rewritten using parallel SIMD instructions, overall speedup about 25%, some commands from Yixin GUI implemented. This version is used in the freestyle group. |
| EMBRYO |
2018 (0.5.4.1847) |
Mira Fontan, Marco Costalba, Joona Kiiski, Gary Linscott, Tord Romstad |
Download |
A lot of errors fixed, speed up, completely rewritten moves generator. Embryo uses part of the search function (highly Gomoku-modified) from Stockfish. |
| EMBRYO |
2017 (0.4.4.1173) |
Mira Fontan |
Download |
The AI is written in C++14, alpha-beta with search extensions, nullmove heuristics, and history heuristics. |
| YIXIN |
2018 (0.7.13) |
Kai Sun |
Download |
The algorithm has been improved. |
| YIXIN |
2017 (0.6.61) |
Kai Sun |
Download |
The algorithm has been improved; bug fixes. |
| YIXIN |
2016 (0.6.9) |
Kai Sun |
Download |
Improvement of algorithm; support hybrid x86-64 to speed up; bug fixes. |
| YIXIN |
2015 (0.4.35) |
Kai Sun |
Download |
The algorithm has been improved in version 2015, a bug has been fixed. |
| YIXIN |
2014 (0.4.3) |
Kai Sun |
Download |
The algorithm has been improved in version 2014. |
| YIXIN |
2013 (0.2.17C) |
Kai Sun |
Download |
Some bugs have been fixed and the algorithm has been improved. |
| YIXIN |
2012 (0.2.14P) |
Kai Sun |
Download |
Same as in the tournament, but obfuscated (protected). |
| YIXIN |
2011 (0.2.7) |
Kai Sun |
Download |
|
| PENTAZEN |
2021 (0.4.18) |
Yuliang Sun |
Download |
Searching and evaluation are optimized; support SWAP-2, multithreading, and some Yixin-Board commands. This version is used in the freestyle and fastgame leagues |
| PENTAZEN |
2021 (0.4.18) |
Yuliang Sun |
Download |
Searching and evaluation are optimized; support SWAP-2, multithreading, and some Yixin-Board commands. This version is used in the standard, renju, and experimental leagues |
| PENTAZEN |
2020 (0.3.2) |
Yuliang Sun |
Download |
Rewrite almost every part. |
| PENTAZEN |
2019 |
Yuliang Sun |
Download |
PentaZen is written in C and uses classic searching techniques. PentaZen is open source. The source code is on https://github.com/sun-yuliang/PentaZen-Gomocup. |
| VIBEFIVE |
2026 (1.0) |
Lu-Chuan Liu |
Download |
VibeFive is a C++ program based on alpha-beta pruning search and a traditional evaluation table. Its development process combines automated AI programming with manual modifications. An LLM agent loop is used for algorithm iteration and parameter tuning. The LLM involved in the development include, but are not limited to, GPT-5.5, Claude opus 4.7, and GLM-5.1. |
| STARDUST |
2019 (1.2) |
Debing Zhang, Ming Zhang, Yumeng Wang |
Download |
StarDust is deep Learning-based AI. |
| XOXO |
2021 (9) |
Jakub Horak |
Download |
Improved threat detection; auto tuning. This version is used in the standard league. |
| XOXO |
2021 (8.0.2) |
Jakub Horak |
Download |
Improved threat detection; auto tuning. This version is used in the freestyle and fastgame leagues. |
| XOXO |
2020 (8.0.1) |
Jakub Horak |
Download |
Better threat detection, support for standard rule, algorithm optimization, some alpha-beta enhancements. |
| XOXO |
2019 (7.5) |
Jakub Horak |
Download |
Completely rewritten, optimized (more than 100x speed up), evaluation and move generation improvement. |
| XOXO |
2018 (5.5) |
Jakub Horak |
Download |
A bug is fixed and the algorithm is a bit improved. This version is used in the standard group. |
| XOXO |
2018 (5.5) |
Jakub Horak |
Download |
A bug is fixed and the algorithm is a bit improved. This version is used in the freestyle group. |
| XOXO |
2017 (5.4.2) |
Jakub Horak |
Download |
Completely rewritten; the heuristic function is improved; support hybrid x86-64 to speed up. |
| XOXO |
2016 (3.0.6) |
Jakub Horak |
Download |
Completely new. |
| XOXO |
2015 (1.12) |
Jakub Horak |
Download |
|
| WINE |
2018 (2.1) |
Jinjie Wang |
Download |
Algorithm performance is improved. |
| WINE |
2017 |
Jinjie Wang |
Download |
|
| SLOWRENJU |
2020 (5.1.4) |
Tianyi Hao |
Download |
A support for SWAP-2 is added. |
| SLOWRENJU |
2019 (5.1.3) |
Tianyi Hao |
Download |
Tuning the parameters with self-playing; supporting the config file. |
| SLOWRENJU |
2018 (5.1.2) |
Tianyi Hao |
Download |
There are some small modifications on the evaluation, although not sure whether there is really a improvement. |
| SLOWRENJU |
2017 (5.1.1) |
Tianyi Hao |
Download |
A little modification on the algorithm; a bug of false-positive forbidden moves in renju rule has been fixed; faster than previous version. |
| SLOWRENJU |
2016 (5.0.11) |
Tianyi Hao |
Download |
Improvement of algorithm and evaluation; support hybrid x86-64 to speed up; add support to renju rule. |
| SLOWRENJU |
2015 (4.1) |
Tianyi Hao |
Download |
Improvement of searching algorithms and parameters. |
| SLOWRENJU |
2014 (4.0) |
Tianyi Hao |
Download |
There is a little improvement with the algorithm. |
| SLOWRENJU |
2013 (3.1) |
Tianyi Hao |
Download |
|
| EGOZERO |
2025 (1.0) |
Yan Tang |
Download |
This version is based on MCTS and ResNet. |
| TITO |
2014 (1.3.2) |
Andrej Tokarjev |
Download |
The bug involed crashing on fast computers, especially Windows 8, has been fixed. |
| TITO |
2010 (1.3) |
Andrej Tokarjev |
Download |
|
| TITO |
2008 (1.2) |
Andrej Tokarjev |
Download |
|
| TITO |
2007 (1.1) |
Andrej Tokarjev |
Download |
|
| TITO |
2006 (1.0) |
Andrej Tokarjev |
Download |
|
| WHOSE |
2019 (20190401) |
Jiawei Huang |
Download |
Add extension to leaf. Add x64 version. |
| WHOSE |
2018 (20180410) |
Jiawei Huang |
Download |
Fix the bug of takeback; Give up the calculation of VCF; Change the data structure; Optimize the algorithm. |
| WHOSE |
2017 (20170412) |
Jiawei Huang |
Download |
|
| DEEPFIRE |
2019 (2.0) |
Shuai Pan |
Download |
Rebuild code and improve speed. Add support for standard rule. Fix some bugs. |
| DEEPFIRE |
2018 (1.0) |
Shuai Pan |
Download |
|
| HEWER |
2018 (4.0.11.540) |
Tomas Kubes |
Download |
There were some changes in move-generator and minor change in evaluation function. Classic Alpha-Beta search with hash table and time management remains unchanged. This version is used in the standard group for Gomocup 2018. |
| HEWER |
2018 (4.0.11.424) |
Tomas Kubes |
Download |
There were some changes in move-generator and minor change in evaluation function. Classic Alpha-Beta search with hash table and time management remains unchanged. This version is used in the freestyle group for Gomocup 2018. |
| HEWER |
2015 (3.6) |
Tomas Kubes |
Download |
It is faster then Hewer 2014, slightly improved, search consider more moves, especially more tricky moves. |
| HEWER |
2014 (3.5) |
Tomas Kubes |
Download |
It is faster then Hewer 2012. Requires .NET Framework 2.0 to be installed. |
| HEWER |
2012 (3.4) |
Tomas Kubes |
Download |
|
| HEWER |
2010 (3.3) |
Tomas Kubes |
Download |
|
| HEWER |
2009 (3.2) |
Tomas Kubes |
Download |
|
| HEWER |
2008 (3.1) |
Tomas Kubes |
Download |
|
| HEWER |
2008 (3.0) |
Tomas Kubes |
Download |
|
| HEWER |
2004 (2.4) |
Tomas Kubes |
Download |
|
| HEWER |
2003 (0.274) |
Tomas Kubes |
Download |
Pn-search algorithm is used. Brief documentation included. |
| GOFIVE |
2018 (0.7.0.0) |
Xin Huang |
Download |
|
| CARBON |
2017 (2.4) |
Michal Czardybon |
Download |
A bug is fixed. |
| CARBON |
2016 (2.3) |
Michal Czardybon |
Download |
An open-source Gomoku program. The AI algorithm was one of the strongest as for 2002 and it's reasonable even now. It is ported to Gomocup interface and further improved by Petr Lastovicka. Carbon supports freestyle, standard, and renju rules. |
| SWINE |
2017 (3.5) |
Jiri Fontan |
Download |
Tuned a little bit; support hybrid x86-64 to speed up. |
| SWINE |
2015 (3.4) |
Jiri Fontan |
Download |
More correct algorithm. |
| SWINE |
2014 (3.2) |
Jiri Fontan |
Download |
Slightly different version then swine 2013. |
| SWINE |
2013 (3.3) |
Jiri Fontan |
Download |
Slightly different version then swine 2012, plays more correctly. |
| SWINE |
2011 (3.2) |
Jiri Fontan |
Download |
It is not exactly the same version used in the tournament, there is fixed bug to support better human playing. |
| SWINE |
2010 (3.1) |
Jiri Fontan |
Download |
|
| SWINE |
2009 (3.0) |
Jiri Fontan |
Download |
|
| SWINE |
2008 |
Jiri Fontan |
Download |
|
| SWINE |
2007 |
Jiri Fontan |
Download |
|
| SWINE |
2005 |
Jiri Fontan |
Download |
|
| SWINE |
2004 |
Jiri Fontan |
Download |
|
| SWINE |
2003 |
Jiri Fontan |
Download |
|
| ARBILLO |
2022 (1.5) |
Lorenzo Piazzo |
Download |
First attempt. Standard (alpha-beta) player. Arbillo supports freestyle rule. |
| ONIX |
2016 (2.0.1B) |
Istvan Virag, Janos Wagner |
Download |
A stable version of Onix 2010. |
| ONIX |
2010 (2.0.1) |
Istvan Virag, Janos Wagner |
Download |
This version sometimes causes a crash. |
| ONIX |
2008 (2.0) |
Istvan Virag, Janos Wagner |
Download |
|
| ONIX |
2007 (2.0) |
Istvan Virag, Janos Wagner |
Download |
|
| SQUIRREL |
2024 (2) |
Xiyuan Liang |
Download |
Algorithm is improved. Some bugs of defense are fixed. Add support for standard and caro rules. |
| SQUIRREL |
2023 (1) |
Xiyuan Liang |
Download |
A simple AI written in C++ with Alpha-beta pruning, iterative deepening search, VCF/VCT search and translation table. |
| GMOTOR |
2011 |
Roman Vancura |
Download |
|
| GMOTOR |
2007 (4.0.2) |
Roman Vancura |
Download |
|
| GMOTOR |
2006 (3.1.0) |
Roman Vancura |
Download |
|
| GMOTOR |
2005 (2.0) |
Roman Vancura |
Download |
|
| PELA |
2023 (7.9) |
Petr Lastovicka |
Download |
Caro rule was added in this version, algorithms are same as in previous version. |
| PELA |
2017 (7.8) |
Petr Lastovicka |
Download |
A bug is fixed. |
| PELA |
2016 (7.7) |
Petr Lastovicka |
Download |
Merge the modifications of AcaPela into Pela; add support to renju rule. |
| PELA |
2015 (ACA) |
Farmer Lu |
Download |
Modification of Pela AI written by Petr Lastovicka. |
| PELA |
2006 (7.5) |
Petr Lastovicka |
Download |
|
| PELA |
2005 (7) |
Petr Lastovicka |
Download |
|
| CHIS |
2017 (4.5) |
Peiming Zheng |
Download |
Speed up. |
| CHIS |
2016 |
Peiming Zheng |
Download |
An open-source Gomoku program. |
| KANEC |
2026 |
Jaroslav Babak |
Download |
Based on proof numbers. Some improvements to mitigate the drawbacks of this method. |
| KANEC |
2009 |
Jaroslav Babak |
Download |
|
| KANEC |
2008 |
Jaroslav Babak |
Download |
|
| KANEC |
2006 |
Jaroslav Babak |
Download |
|
| KANEC |
2005 |
Jaroslav Babak |
Download |
|
| ZETOR |
2017 (4) |
Tomas Chmela |
Download |
Contains some improvements and bugfix. |
| ZETOR |
2016 (3) |
Tomas Chmela |
Download |
Partially rewritten. |
| ZETOR |
2015 (1) |
Tomas Chmela |
Download |
|
| ZETOR |
2014 (0.7) |
Tomas Chmela |
Download |
AI is based on alpha-beta pruning of minimax agorithm with simple hash-table. Search depth at fast game is 2 half-move. Board size is supported from 15x15 to 20x20. Rule of win is selectable - exactly five wins or five and more wins. Program is written in C. |
| ANGEL |
2026 |
Yuan Xie |
Download |
Starting from this version, an AlphaGo Zero-like algorithm is adopted, replicating the four optimization points from Chapter 3 of the KataGo paper. |
| ANGEL |
2022 |
Yuan Xie |
Download |
Negamax Alpha-Beta search, possible position sort, quick believable VCF search, Iterative Deepening. Angel supports freestyle rule. |
| EULRING |
2016 (3.0) |
Junyi Bian |
Download |
|
| PECUCHET |
2007 |
Vilem Walter |
Download |
|
| SPARKLE |
2017 (2.2) |
Jaechan Lee |
Download |
Slightly improved from the previous AI Ignitor. |
| JUDE |
2007 (1.0) |
Tian |
Download |
|
| MONKEY |
2019 (1.1) |
Letian Yuan |
Download |
The Monkey is a very simple C# program which is mainly based on Alpha-Beta pruning algorithm. It can calculate VCF of 100 steps accurately. |
| DDQKCONQUER |
2026 (3.3) |
Shunyi Sui |
Download |
DDQK-Conquer has greatly improved the computing speed, effectively combining C++ and Python, so that the space and speed can be improved and optimized more effectively. It is no longer limited to the traditional alpha-beta pruning algorithm, but further adopts an algorithm similar to NNUE. |
| DDQKCONQUER |
2022 (1.0) |
Shunyi Sui |
Download |
Mainly using the use-value network algorithm, alpha beta screening is also involved. DDQK-Conquer supports freestyle rule. |
| PUREGM |
2021 (2.0.0) |
Paul Dwyer |
Download |
MCTS algorithm guided by a trained ANN. |
| PUREGM |
2020 (1.0.2) |
Paul Dwyer |
Download |
The pattern matching algorithm remains similar to previous versions of PureGM submitted years ago but is now the heuristic for an MCTS algorithm. |
| PUREGM |
2013 |
Paul Dwyer |
Download |
Iteratively deepened Alpha-Beta pruned game tree search with template based heuristic. There is some tweaking and bug fixing compare to version 2012. |
| PUREGM |
2012 |
Paul Dwyer |
Download |
Iteratively deepened Alpha-Beta pruned game tree search with template based heuristic. It is not exactly the version used in the tournament. |
| PUREGM |
2011 |
Paul Dwyer |
Download |
Iteratively deepened Alpha-Beta pruned game tree search with template based heuristic. It is not exactly the version used in the tournament, there is an extra fixed bug to enable human playing. |
| PUREGM |
2009 |
Paul Dwyer |
Download |
|
| PUREGM |
2008 |
Paul Dwyer |
Download |
|
| MENTAT |
2005 (Q1.6) |
Vojta Havranek |
Download |
Updated QMentat for Gomocup 2005. The algorithm is based on minimax. |
| MENTAT |
2003 (Q1.4) |
Vojta Havranek |
Download |
Ancestor of Xmentat. |
| MENTAT |
2002 (X1.9) |
Vojta Havranek |
Download |
|
| MENTAT |
2001 (1.8) |
Vojta Havranek |
Download |
|
| DJALL |
2018 (1.6) |
Ladislav Petr |
Download |
|
| DJALL |
2017 (1.1) |
Ladislav Petr |
Download |
|
| NOESIS |
2004 |
Mark Mammel |
Download |
|
| IMRO |
2015 (1.2) |
Imrich Kovac |
Download |
|
| IMRO |
2013 (1.0) |
Imrich Kovac |
Download |
|
| FASTGOMOKU |
2015 (20150225) |
Zheng Wang |
Download |
|
| FASTGOMOKU |
2014 (20140209) |
Zheng Wang |
Download |
FastGomoku 2014 was used instead of FastGomoku 2015 in Gomocup 2016 because it has been shown that version 2014 is significantly stronger than version 2015. |
| PUSKVOREC |
2003 |
Patrik Bezdicek |
Download |
|
| ZHOUMIAN |
2022 (0.1) |
Shangren Shang |
Download |
A low-IQ AI based on C language. Zhoumian supports freestyle rule. |
| LICHT |
2014 (1.0) |
Xinhui Li |
Download |
|
| PISQ |
2004 (7) |
Martin Petricek |
Download |
A slightly upgraded Pisq from tournament 2004. |
| PISQ |
2003 (6) |
Martin Petricek |
Download |
Windows version of pisq5 and pisq6 from Gomocup 2003. |
| PISQ |
2000 (3/4/5) |
Martin Petricek |
Download |
You can set up it's difficulty by renaming from pisq1.exe to pisq5.exe. Level 5 is default. |
| LIGHTBLUE |
2018 (2.4) |
Erwan Simon |
Download |
|
| TKGOMOKU |
2026 (0.4) |
Tae Kim |
Download |
It uses iterative deepening alpha-beta/PVS-style negamax search, transposition table, threat-aware move ordering, immediate tactical detection, VCF/VCT-style probes, handcrafted line/pattern evaluation, and lightweight NNUE/policy/opening-book files. |
| TPTC |
2025 (1.0) |
Bruno Yerro |
Download |
TPTC is written in python. |
| VALKYRIE |
2013 (1.0) |
Kokubun Takashi |
Download |
Gomoku articial inteligence written in five days, Valkyrie is opensource project and it is hosted on github. |
| CRUSHER |
2010 |
Nils Beckmann |
Download |
Written in Delphi. Based on evaluation function withou search engine. |
| STAHLFAUST |
2016 (1.0) |
Marco Kunze, Sebastian Nowozin |
Download |
An open-source gomoku program. It is ported to Gomocup interface by Petr Lastovicka. |
| FIVEROW |
2008 |
Yaroslav |
Download |
|
| PROLOG |
2014 (0.1) |
David Macek |
Download |
|
| AIBB |
2020 (0.0.1) |
Cuihong Jiang |
Download |
AiBB is a basic engine using alphabeta algorithm for searching. |
| MUSHROOM |
2011 (3.14159) |
Kai Chen |
Download |
|
| WHATATEAM |
2018 (0.1) |
Fanny Tavart, Oscar Stefanini, Ronan Boiteau |
Download |
|
| LAVENDER |
2023 |
Hoang Vinh Nguyen |
Download |
This version only supports Standard Caro rule. |
| STARPOINT |
2026 (2026-freestyle-15) |
You Huang |
Download |
Starpoint is a Gomoku engine based on Monte Carlo Tree Search and transformer-based evaluation, with Proof-Number Search used for VCT search inside the engine. |
| SKYZERO |
2026 (5.1) |
Sky Cheng |
Download |
SkyZero is a Gomoku AI powered by the Gumbel AlphaZero algorithm. |
| OLECHKA |
2026 (1) |
Dima Suraev |
Download |
alpha-Beta pruning search with Transposition Table, Zobrist Hashing, Iterative Deepening and partial depth caching. Custom bitwise evaluation optimized for Freestyle rules. Secret heuristics. |
| FIGRID |
2026 (0.7.1) |
Hogyung Choi, Bobo Wu |
Download |
α-β engine with NNUE evaluation (via the noru Rust library). Standard search techniques (IID, aspiration, IIR, LMR, TT, qsearch) plus a Pattern4-mini threat cache and continuation history driving move ordering. NNUE evaluation via the noru library. Adds phase-based time management. |
| FIGRID |
2025 (0.3.0) |
Bobo Wu |
Download |
First version, really weak. |
| MONAS |
2026 (1.0) |
Guanwu Liu |
Download |
Monas employs a pre-computed pattern matching table for fast threat recognition, combined with Alpha-Beta search; a fallback mechanism is invoked when needed to ensure stable move generation. |
| ALPHAGOMOKU(GS) |
2018 (1.0) |
Gerry Sun |
Download |
AlphaGomoku is based purely on deep neural network. |
| CHLORIS |
2026 |
Nelly Semenova |
Download |
Tournament engine for Standard-15 Gomoku with a custom NNUE evaluation trained for Standard 15x15 play on approximately 3 million positions, iterative-deepening alpha-beta search, and a dedicated tactical layer, including tactical predicates, forcing-move checks, and urgent-threat defense. The combined positional evaluation was tuned through games against stronger engines. |
| MINTAKA |
2026 (pre-alpha) |
JeongHyeon Choi |
Download |
Renju engine written in Rust, currently under development, based on PVS and (temporary) heuristic algorithm. |
| XL |
2016 |
Chengtao Chen |
Download |
An open-source Renju program. It is ported to Gomocup interface by Kai Sun and Tianyi Hao. |
| AI |
2002 (1000) |
Jan Macura |
Download |
|
| AI |
2002 (620) |
Jan Macura |
Download |
|
| AI |
2001 (016) |
Jan Macura |
Download |
|
| AMOEBA |
2010 |
Galli Zoltan |
Download |
The program has three main components: an opening book for the early game, a Monte-Carlo tree search algorithm, a heuristic evaluation function tuned by the cross-entropy optimization algorithm. It was not used in the tournament since 2016 because its memory usage exceeds the quota too often on Windows 10. |
| AMOEBA |
2009 (5.3) |
Galli Zoltan |
Download |
The program has three main components: an opening book for the early game, a Monte-Carlo tree search algorithm, a heuristic evaluation function tuned by the cross-entropy optimization algorithm. |
| BENJAMIN |
2003 |
Benjamin Vejnar |
Download |
This AI is inclined to draw, a match Benjamin vs. long thinking AI takes a lot of time. |
| BESTMOVE |
2004 |
Xiangdong Wen |
Download |
First brain we received from abroad!(USA) Uses new protocol. Tournament 2004 version. |
| BETA |
2005 (1.0) |
Tomas Kubes |
Download |
|
| BRUTEFOR |
2002 |
Vasek Synacek |
Download |
|
| CURNPIS |
2000 |
Jan Curn |
Download |
|
| DEEPBLUE |
2000 |
Vojta Havranek |
Download |
|
| DOLORES |
2005 |
Tomas Kubes |
Download |
|
| EGO |
2021 (8.0) |
Yan Tang |
Download |
Bug fixed. |
| EGO |
2019 (7.0) |
Yan Tang |
Download |
Update the function of evaluation. |
| EGO |
2018 (6.0) |
Yan Tang |
Download |
Ego is written in C++. |
| GAMMAKU |
2017 (1.0) |
Zhihao Zhou, Ziyi Dou |
Download |
GammaKu for fastgame group. |
| GAMMAKU |
2017 (1.0) |
Zhihao Zhou, Ziyi Dou |
Download |
GammaKu for freestyle group. |
| GORO |
2018 |
Viktor Barykin |
Download |
Some weights were changed. |
| GORO |
2017 |
Viktor Barykin |
Download |
Some parameters were changed. A protection makes it expired after some time (it is sending emails). |
| GORO |
2016 |
Viktor Barykin |
Download |
Changed a bit. A protection makes it expired after some time (it is sending emails). |
| GORO |
2015 |
Viktor Barykin |
Download |
Quite different. A protection makes it expired after some time (it is sending emails). |
| GORO |
2014 |
Viktor Barykin |
Download |
A protection makes it expired after some time (it is sending emails). |
| GORO |
2013 |
Viktor Barykin |
Download |
A protection makes it expired after some time (it is sending emails). |
| GORO |
2012 |
Viktor Barykin |
Download |
A protection makes it expired after some time (it is sending emails). |
| GORO |
2011 |
Viktor Barykin |
Download |
A protection makes it expired after some time (it is sending emails). |
| GORO |
2010 |
Viktor Barykin |
Download |
A protection makes it expired after some time (it is sending emails). |
| GORO |
2009 |
Viktor Barykin |
Download |
A protection makes it expired after some time (it is sending emails). |
| GORO |
2007 |
Viktor Barykin |
Download |
A protection makes it expired after some time (it is sending emails). |
| GORO |
2005 |
Viktor Barykin |
Download |
|
| GORO |
2004 |
Viktor Barykin |
Download |
|
| HGARDEN |
2007 (1.3) |
Bingqing Han |
Download |
|
| HGARDEN |
2006 (0.4) |
Bingqing Han |
Download |
It is the version used in the 1st tournament AI vs. Human (11/2006). |
| HGARDEN |
2006 (0.4) |
Bingqing Han |
Download |
It is the version used in the Gomocup 2006. |
| HGARDEN |
2005 (0.2) |
Bingqing Han |
Download |
|
| HIGHFIVE |
2014 (0.7) |
Peter Karpov |
Download |
It was not used in the tournament since 2016 because it crashes on Windows 10. |
| IGNITOR |
2016 (1.0) |
Jaechan Lee |
Download |
|
| KRYSA |
2002 |
Jiri Fontan |
Download |
Kerry is written in Delphi, so it can run under the windows only. |
| LEONARDO |
2001 |
Milan Matlak |
Download |
|
| NABAMOKU |
2008 |
Jiri Isa |
Download |
|
| NABAMOKU |
2007 |
Jiri Isa |
Download |
|
| PENTAZENNN |
2022 (0.5.0) |
Yuliang Sun and contributors of Katago |
Download |
Switch to MCTS with NN evaluation modified from KataGo. The alpha-beta search serves as a parallel surewin finder now. PentaZenNN supports standard and renju rules. This version is compiled with AVX2 instructions. |
| PENTAZENNN |
2022 (0.5.0) |
Yuliang Sun and contributors of Katago |
Download |
Switch to MCTS with NN evaluation modified from KataGo. The alpha-beta search serves as a parallel surewin finder now. PentaZenNN supports standard and renju rules. |
| PENTAZENNN |
2022 (0.5.0) |
Yuliang Sun and contributors of Katago |
Download |
Switch to MCTS with NN evaluation modified from KataGo. The alpha-beta search serves as a parallel surewin finder now. This version is for the unlimited league. |
| PUREROCKY |
2016 |
Paul Dwyer |
Download |
Most of the tree based algorithms are removed from the previous AI PureGM, and the AI has been trained over the last several months to get stronger. |
| RENJUSOLVER |
2011 (2008) |
Xiangdong Wen |
Download |
Based on Renju VCT Problems Solver, parameters were tuned up by XR. The configuration file is set to play standard rule. |
| RENJUSOLVER |
2011 (2008) |
Xiangdong Wen |
Download |
Based on Renju VCT Problems Solver, parameters were tuned up by XR. The configuration file is set to play renju rule. |
| RENJUSOLVER |
2011 (2008) |
Xiangdong Wen |
Download |
Based on Renju VCT Problems Solver, parameters were tuned up by XR. The configuration file is set to play freestyle rule. |
| TIANSHU |
2020 (1.1) |
Guang Yang |
Download |
Tianshu is written in Python. The VCT/VCF algorithm is good. |
| TRUNKAT |
2002 |
Jiri Trunkat |
Download |
AI Trunkat, the winner of Gomocup 2002. You can set up it's difficulty by renaming from trunkat1.exe to trunkat5.exe. A file CSWDPMI.EXE is required to be in the same directory with the brain. The program si written in DOS and it is not xompatible with Windows 8 and higher. |
| WULIN |
2019 (1.0.0) |
Chao Liu |
Download |
Wulin is based on minimax search with alpha-beta pruning and a special VCF/VCT algorithm. (Comment from organizers: We notice that Wulin stops working after the tournaments, which we guess may be due to some protection that makes it expired after some time.) |
