C-Score discrepancy

Hi,

I haven't fully understood on what basis the models predicted by I-TASSER is categorised on.

The confidence is measured by the C-Score as stated, so I had a query in one of the sequences I'd sent to I-TASSER for prediction. In the image attached, the C-Score for Model1 is -1.76 whereas the C-Score for Model2 in 1.72. As the Model2 has a higher C-Score, do I need to choose Model2 as the best-predicted model by I-TASSER? Even the C-Score values differ by a very small number, there is consdierable difference in the structure between both the models.

Could someone help me out on this?

Thank you for your time!
Ahish M Sujay