Accurate phase diagram modeling is foundational for the design of materials with targeted properties. The CALPHAD (CALculation of PHAse Diagrams) methodology provides a powerful framework for thermodynamic modeling across multicomponent systems. However, the reliability of CALPHAD predictions depends critically on the quality and consistency of the underlying Gibbs energy descriptions, particularly for pure elements in different crystal structures. A key input in CALPHAD modeling is the lattice stability—defined as the energy difference between two competing crystal structures (e.g., fcc vs bcc) of an element. These values serve as essential reference points for constructing Gibbs energies of solution phases. Unfortunately, many of these structural forms are thermodynamically unstable at ambient conditions, making direct experimental determination of their energies infeasible. Historically, CALPHAD databases (e.g., SGTE) have relied on empirical or semi-empirical estimates of lattice stabilities, which can introduce significant uncertainty. These estimates often lack internal consistency and are limited in scope, particularly for less-studied elements or unstable phases. This limits the predictive capability of CALPHAD assessments and poses a challenge when modeling systems involving metastable or transient phases. To address this limitation, density functional theory (DFT) calculations are commonly applied to determine the total energies of elements in a range of crystal structures, including mechanically and dynamically unstable phases. However, naively modelling unstable phases at their ideal positions can lead to significant inconsistencies with the CALPHAD models. More involved methods are needed to get more realistic energies for the lattice stability evaluations. One such method is the inflection-detection method, where unstable phases are assigned the energy at the inflection point on the path of transformation. Another way is to compute and extrapolate finite temperature results, as has been done with the direct-upsampling method. In this project, we will use these methods to compute lattice stabilities for metastable and unstable phases. A key objective is to compare the performance of the different methods. More importantly, the computed lattice stabilities will be used to improve and extend third generation CALPHAD databases.