No one has gone nuts, models are imperfect. Keep your pants on.
To Neven's point, models (and analysts) can be fooled.
For example, consider the three images I'm posting here.
The first is a snippet from this:
http://arctic.atmos.uiuc.edu/cryosphere/NEWIMAGES/arctic.seaice.color.000.png... which details much of the Chukchi and northern Beaufort north of the Bering and NE of Wrangel Island.
It looks very solid and is exactly the sort of thing which is generating fairly high extent numbers. There are a lot of areas around the central arctic that look like this.
Next is a screen grab from EOSDIS Worldview, for 2015-07-03 - current conditions in the arctic. You can see from it exactly why CT thinks the region is both well iced, and high concentration.
The last is a zoom in to the center of that EOSDIS shot, an area of approximately 10000 (ten thousand) square kilometers. I think many of you will agree, the ice in that image is sick. There is no contiguous floe in that image of more than 100KM2, and only one of those. Judging heuristically, the modal flow size is well under 1KM2, and the median about that. Much of the ice in that image is broken up so small (under 300 meters), that lateral melt is potentially a factor in how fast it will melt.
Now, I'm not sure what exactly the weather is going do to this ice, but I think I can conclude that the satellite image suggests much less robust conditions than one might take away from either the extent or area numbers, or the image presented by the model.