OpenAI's recently launched reasoning AI model, o1, has sparked intrigue as users report it occasionally processes thoughts in languages like Chinese or Persian, even when prompted in English. While the final responses remain in English, the intermediate reasoning steps sometimes occur in other languages. Experts suggest this could be due to the model's training data, which includes diverse linguistic sources. Other theories propose that the model's language switches reflect efficient problem-solving strategies. OpenAI has not commented on this phenomenon, leaving experts to speculate on the implications of such behavior.