I gave Gemini 2.5 Pro the same prompt and it did much better, keeping track of the pot and stacks correctly and having the players make reasonable decisions. (Probably not optimal, but they shouldn't be given the scenario). The only mistake I saw was saying that one player had an open-ended straight draw when it was actually a gutshot.

On the other, asking the LLM to generate the source and test code for a poker game would be more productive. Could actually be a better player than some people with an AI algorithm for decison making.