From a new paper in Geographical Review by Walsh et al, and a press release from the University of Colorado. Note this is ice extent, which is more than just sea ice.
I'm looking to see if there's a time series plot....
Added: Here it is:
For comparison, the minimum sea ice extent in September 2012 was 3.62 Mkm2 (average for the entire month), so the ice extent was almost all sea ice at that minimum.