Long polymethines are well-known experimentally to symmetry-break, which dramatically modifies their linear and nonlinear optical properties. Computational modeling could be very useful to provide insight into the symmetry-breaking process, which is not readily available experimentally; however, accurately predicting the crossover point from symmetric to symmetry-broken structures has proven challenging. Here, we benchmark the accuracy of several DFT approaches relative to CCSD(T) geometries. In particular, we compare analogous hybrid and long-range corrected (LRC) functionals to clearly show the influence of the functional exchange term. Although both hybrid and LRC functionals can be tuned to reproduce the CCSD(T) geometries, the LRC functionals are better performing at reproducing the geometry evolution with chain length and provide a finite upper limit for the gas-phase crossover point; these methods also provide good agreement with the experimental crossover points for more complex polymethines in polar solvents. Using an approach based on LRC functionals, a reduction in the crossover length is found with increasing medium dielectric constant, which is related to localization of the excess charge on the end groups. Symmetry-breaking is associated with the appearance of an imaginary frequency of b2 symmetry involving a large change in the degree of bond-length alternation. Examination of the IR spectra show that short, isolated streptocyanines have a mode at ~1200 cm-1 involving a large change in bond-length alternation; as the polymethine length or the medium dielectric increases, the frequency of this mode decreases before becoming imaginary at the crossover point.