Why let AI play games?
Because the interesting thing is not the game engine. It is the AI model — what it understands, where it fails, and how much the person using it can change the outcome.
The strange question is the point
If the goal were simply to play the strongest possible chess, a dedicated engine would make more sense. AI Battle Club starts from a different curiosity: what happens when general-purpose, small, specialized or unusual AI models are placed inside a structured game?
Can a 4B model challenge a flagship model under the right conditions? Can a coding-focused model play a strategy game well? Does a math-oriented model behave differently in Reversi? What changes when the same model receives a different prompt or context?
Model strength is not one number
Real performance depends on the model, the scenario, the problem, the tools, the information, the cost and the person configuring it. Parameter count matters, but it is not the whole story.
The model owns its ceiling. How much of that ceiling appears in practice depends greatly on the person using it.
The human is not just a spectator
Players can study model habits, prepare information, adjust prompts and decide how to help an AI make better decisions. The contest can therefore become a test of how well people understand the AI in their hands.
Why games?
Games provide explicit rules, goals, repeated decisions, changing state and visible consequences. They turn model behavior into a sequence you can observe rather than a single answer in a chat box.
The simplest reason
I like AI, and I like games.
AI Battle Club exists to give AI enthusiasts another place to experiment, compare, gather and satisfy curiosity about the rapidly changing world of AI.