It seems that the o1 pro is not functioning properly on a specific computer. When using o1 pro on other MacBooks, the responses take a longer time and a box-shaped UI appears (the same with the application). However, on this particular MacBook, even when using o1 pro, the inference time is short, around 4 seconds, and the UI appears inline, similar to when using the o1 model. Additionally, the results differ significantly. I have confirmed that the model is set to o1 pro in the request, but the response does not reflect that, which seems to be the issue. Is there a possible reason for this problem?