5. The full source code and the models used

Purpose of this post

This post publishes the source code used for the Othello AI. The code is loaded only when you open a file.

Models used

• Gemma 4 E4B (mlx-community/gemma-4-e4b-it-4bit): the default model of the VLM server. It currently fails to load and is not used.
• mlx-vlm: runs the model on Apple Silicon.
• Local AI: not a trained model. The source has alpha-beta search; the deployed web has the one-move weighted AI.
• There is no trained model yet.

Folder layout

• src: Qt6 C++ app. Rule engine (othello_board), search AI (aiplayer), VLM client (tcpvlmclient)
• vlm_server: OT88 TCP server, rules, Gemma 4 download
• othello-web: Next.js web play. Rule engine (board.ts), search AI (ai.ts), game storage (db.ts), VLM relay (api/vlm/move, ot88Client.ts)

Server deployment settings, monitoring and automation scripts, and game data and logs are excluded from the release. Web operation code such as ads and the list of other games, and configuration files are also left out, so some imports point to unpublished files.

Full source