This is quite reasonable. Because after NTO transformation, the eigenvalue of the orbitals is no longer orbital energy but contribution of NTO to the selected electronic excitation. An occupied and a unoccupied NTO compose a NTO pair; in each NTO pair the two NTOs share the same eigenvalue.

If you are still confused, please check Section 3.21.6 of Multiwfn manual to gain basic knowledges about NTO method.

I did TDDFT calculations using ORCA and loaded .gbw file to run NTO analysis. Did calculation and got value of HOMO-LUMO energy.

But when I repeat the process again I am not getting energy gap. HOMO and LUMO have same energies

55(H ) --> Charge:  1.000000  x,y,z(Bohr): -11.214523  -3.650781   0.598930
   56(H ) --> Charge:  1.000000  x,y,z(Bohr):  -6.178762   1.802666  -4.288639
   57(H ) --> Charge:  1.000000  x,y,z(Bohr):  -0.057032  -5.519588   2.665837
Note: Orbital   105 is HOMO, energy:    0.978720 a.u.   26.632331 eV
       Orbital   106 is LUMO, energy:    0.978720 a.u.   26.632331 eV
       HOMO-LUMO gap:    0.000000 a.u.    0.000000 eV      0.000000 kJ/mol

What may be the reason?