Different results when *DEFORMATION PLASTICITY vs *PLASTIC used for modelling plasticity of ductile meals in ABAQUS?

I got different results for a tensile test simulation when I used *DEFORMATION PLASTICITY (DP), instead of *PLASTIC (P) to model the same material. Ramberg-Osgood (R-O) parameters were directly provided for DP model while true stress/plastic strain values were calculated using the same R-O parameters for P model. Could somebody provide some insight into why that happened? This is not a fracture problem.

A simulation of tensile test (ASTM E8) of a solid aluminum bar (axisymmetric, reduced integrated elements) under monotonic loading with deformation plasticity and classical metal plasticity models used separately to model the aluminum, also showed the same discrepancy. The results of the tensile test simulation are provided below. Why do these two models, describing the same material response, result in different behaviour of the model?

Perfect agreement between calculated stress/strain curve and actual R-O curve was observed when *PLASTIC (rate independant, isotropic hardening) was used. But using *DEFORMATION PLASTICITY, the model response deviated (from *PLASTIC) after the linear region and reached UTS at very small strains (See figure below).

Initially I used Ramberg-Osgood parameters, E= 70 GPa, Poisson's ratio = 0.33, Yield Stress = 195 MPa, Exponent = 48, Yield offset = 0.002 (0.2% proof)

After I posted this I saw Abaqus require α as the yield offset. So, analysed again with yield offset = 0.71795. The new stress/strain curve resulting from the analysis is below. Now the results do match upto larger strains. Still the discrepancy is there.

Since I used the same R-O parameters, directly for *DEFORMATION PLASTICITY and to calculates true stress/plastic strain for *PLASTIC, I expected the same response from the model. Could somebody provide some insight into the problem? Thanks in advance!