Same 2048, different algorithms

No more swiping. Let the AI play the game for you. Progress through the 2048 game by unlocking more advanced algorithms and machine learning models. Choose from Basic Heuristics, Monte Carlo Simulation, Pure Monte Carlo game search, Deep Q Learning (DQN), and many more to be added.

Training Your Own Model

Pick from one of the pre-trained deep learning models, or train your own machine learning models from scratch. Experience what it is like to be a skillful machine learning / AI engineer. You can also save your model and load it later without losing the progress.

Fine tuning parameters

Many parameters to be adjusted and tuned to your liking in order to get the most out of each algorithm. Customize the scoring function, number of simulations, etc for more than 1000 different combinations to grid-search through.

Upgrade your AI capabilities

Earn chips by playing the game and purchase upgrades to unlock many quality-of-life (QOL) capabilities. Upgrades include faster solve speed, auto-restart, training statistics and many more. Have fun playing as an AFK game.

Get the highest high score

Record your high scores together with the algorithms you used at the end of each game. Use your past performance to tune the AI and achieve new scores. 2048 is not the end, it is just the beginning.

Compete on global leaderboard

Submit your high score to the global leaderboard, compete against other AIs to achieve the ultimate global high score.

Features:

- Original AI simulation game concept
- Choose pre-trained models, such as Deep Q Learning
- Train your own AI models via Deep Q Learning
- Save and load training progress
- Super fast AI solving speed for satisfying solving experience
- Customizable and upgrable AI algorithms and parameters
- Gameplay powered by tf.js (Tensorflow) engine to give an authentic AI experience
- Upgrade AI speed and capabilities for faster and more enjoyable gameplay
- New game from the creator of Black Hole by thACk, Winner of the Best Gameplay Award of NUS GDC 2012