(Solved) H2O PBE J-coupling incorrect


Click here for full thread
Clicked A Few Times
I'm calculating the H-H J-coupling of the water molecule using the following input:
start job
geometry units angstroms
O 0.0  0.0      0.110812
H 0.0  0.783976 -0.443248
H 0.0 -0.783976 -0.443248
end
basis
 O library cc-pvdz
 H library cc-pvdz
end
dft
 xc pbe96
 grid lebedev
 tolerances tight
 odft
 noio
end
property
 spinspin 1 2 2
end
task dft property

Here is the J-coupling and its various terms in comparison to the ones obtained with Dalton and FHI-aims:
|          | Fermi contact | Paramagnetic | Spin-dipole | Diamagnetic |    Total |
| FHI-aims |        0.7096 |       8.1231 |      0.4637 |     -7.8966 |   1.3997 |
| Dalton   |        0.7149 |       8.1222 |      0.4646 |     -7.8967 |   1.4050 |
| NWchem   |     -539.8454 |     -28.6307 |    -14.0512 | 225128.6878 | 224546.2 |

Full output is available here. As you can see, the numbers not only differ, but are clearly unrealistic. I tried with other functionals and with Hartree-Fock but the results are overall similar. Could somebody check the above input and tell me if there is a mistake?