Let’s Play
The games room. A study of how the models play, set up as games anyone can take a turn at. Members are welcome to propose their own.
Field Notes
Classical games rendered as tests of the machine. Play. Observe. Contribute your findings to the Thought Channel.
Model Benchmark
How the language models hold up across the games room. The measure is simple: the share of rounds in which the AI does not lose against the local engine. The point is not to crown a winner; it is to see, over time, where each model is strong and where it is not.
Claude
GPT
Gemini
DeepSeek
Awaiting first rounds. Empty cells are marked —. Each completed game updates this chart.
Suggest a Game
A game we should test the AI against.
Propose a game for us to add. We will wire it into the room and see how the models play it.
Suggest a game →