OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Threes AI

Threes AI is an open-source game-playing project for the Threes! puzzle, combining deck-aware expectimax, N-tuple temporal-difference learning and AlphaZero-style methods. It includes a web demo and a self-hosted game environment intended for model training and public demonstrations.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Threes AI — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when your project or learning goal is specifically about Threes! gameplay, game search or agent-learning experiments, rather than a general-purpose AI model.

Who it is for

  • Developers studying search and learning approaches for board-game agents.
  • Threes! players who want to view the project’s browser demo.
  • Builders who want to explore a game environment with a Go server and Meteor web client.

Documented uses

  • Inspect or adapt a Threes! game-playing agent.
  • Observe the AI through the project’s web demo.
  • Start the self-hosted game locally as an environment for experimentation.
Before you choose

Requirements & limits

The repository is public and specifies a GPL-3.0 license. Its README links to a web live demo.

  • The self-hosted setup uses separate Go server and Meteor web components communicating over WebSockets.
  • The README documents Docker and local startup paths; local web startup requires a configured Meteor environment.
  • The README specifies Node 8.9.4 for running the production web bundle described in its Meteor 1.6 instructions.
  • The separate Python script for controlling the official browser game uses Chrome remote debugging and is noted as prone to WebSocket disconnections.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Run the self-hosted web game locally

Start the documented Go server and Meteor web client for the project’s Threes! game.

Before you start

  • The project repository
  • Go
  • Meteor configured in the local environment

Steps

  1. From the project root, run the command below to start the Go server on port 9000.

    Documented command
    go run main.go

  2. In a second terminal, enter threes-ai/threes! and run meteor.

  3. Open the locally served web client at http://localhost:3000.

Expected outcome

The Go server and web client are running locally as separate components.

Guide limitations

  • The guide covers local startup, not training or deploying a model.
  • The README’s production-bundle instructions specify Node 8.9.4; that version requirement applies to the Meteor 1.6 build described there.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
8 October 2026
Original website title
halfrost/threes-ai · Official homepage title · retained in its source language
How we review sourcesSuggest a correction