Novel codes allow the prediction of parameters with increased resolution, in comparison to the conventional approach, sometimes using meshing smaller than the size of the fuel pin. Optimally, complex geometric structures, like the ones included in the VVER core e.g. corner stiffeners, would be modeled explicitly by novel neutronic codes. In this paper, the performance of the deterministic neutron transport code nTF is studied for VVER geometries. The code is verified standalone with Monte Carlo reference solutions for a range of Hot Zero Power (HZP) models, from 2D pincells to the 3D full core of a VVER-1000, described in the X2 benchmark. Several modeling options and approximations are studied to optimize the nTF X2 3D core model. The updated version of nTF, including all necessary modifications to improve its VVER modeling capabilities, presents good agreement with the reference solutions for the simple VVER configurations and the critical state of the X2 core. Specifically for the full core critical state, the RMS of the pin power difference is 1.5% and the eigenvalue difference 74 pcm with respect to a Monte Carlo reference solution. Any resulting discrepancies are analyzed, the more significant ones with the use of the lattice code CASMO5-VVER that employs similar methods as nTF. The optimized nTF X2 full core model is validated with the experimental data included in the X2 benchmark, produced by HZP start-up tests. The differences of the nTF results with the measured and simulated data of the X2 benchmark remain within the accuracy limits set by the utilities and the literature of deterministic code verification for most cases. The related discrepancies are discussed in detail.