Thanks for your help.
I figured out the problem. I specified a stacking factor of 0.78 in the y direction, which messed up the solution of the B and H field. If I set the stacking factor to 1, the results are a lot closer.
One potential problem was that I used a coarse meshing resolution. Perhaps Ansys Maxwell does not work well if you combine a a coarse meshing resolution, a stacking factor (0.78), and a nonlinear core.
Stacking factor of 0.78 is on the left, stacking factor of 1 is on the right. You can see that with the stacking factor of 0.78, the B field only went vertically up and down, in the direction of laminations.