I only did one move as I did not want to use up my quota.
You can try the same prompt with some of the other text models but the user interaction and flow of the game is much better with the o1 model.
Play a game of chess (model 4o mini)
If I had a higher limit/quota I would try other things like filling out popular forms, asking for help with common problems like fixing a leaky faucet, etc. Would be curious how to set up prompts such that the model works as a fluent chat session but helping you along the way to interact after each step instead as a single response as is done when playing a game of chess.