Go 9x9 -ZX Spectrum-

Go board game "Go 9x9" (2026) for ZX Spectrum with Monte Carlo AI

Go 9x9 is a board game implementation for a home computer, featuring a 9x9 variant of the Go board game. The artificial intelligence uses a Monte Carlo tree search combined with various optimizations, including RAVE, local tactical analysis, and pattern-based evaluation. Pattern recognition is trained on a large set of positions evaluated by KataGo. The AI performs around 150 simulated games before each move and uses the outcomes to choose the next action.

The game includes an opening library of 300 positions, features for area scoring with komi, positional superko, and supports two-player mode. It provides options for setting move duration from 10 to 60 seconds and handicaps from 2 to 5 stones. Dead stones are estimated, territory is displayed, and final score is counted at the end. All functionality is implemented in Z80 assembly within a 48K memory constraint. 

This program was developed by LAWLORD and released in 2026.

Source:https://zxart.ee/prod/604970