In this paper we present the results of ab initio model potential (AIMP) embedded-cluster calculations on the ground (4)A(2g) and excited T-4(2g) State levels of V2+-doped KMgF3, KZnF3, KCdF3, and CsCaF3. Complete active space SCF (CASSCF) and averaged coupled-pair functional (ACPF) calculations are performed on the (VF6)(4-) embedded cluster. The AIMP embedding potentials represent both static and relaxed/polarized lattice effects which are found to be an important refinement due to the large local distortions produced by the V2+ impurity. The calculated local distortions are found to be considerably large, but much smaller than expected in terms of the mismatch of ionic radius of the impurity and the substituted cation. The host dependency of the crystal field splitting, which was found to violate the simple ligand field R(-5)-law in a wide family of V2+-doped halide crystals, if R is the metal-ligand distance in the host, is also examined and is found to be quite close to the simple ligand field theory prediction, as long as the calculated impurity-ligand distances are considered. The results of the (4)A(2g) --> T-4(2g) absorptions are in close agreement with available experimental data. The comparatively high values of the calculated T-4(2g) --> (4)A(2g) fluorescence indicate the need of the inclusion of intracluster Jahn-Teller coupling of the excited state. However, the host variation of fluorescence is, again, well reproduced.