Grain boundary properties such as free energy, strength or mobility are of interest for understanding and predicting the deformation mechanisms of polycrystalline materials. In particular, the effect of temperature on these properties is of high interest, as it is an important factor in many manufacturing and in-service processes. In the computational setting, these properties are commonly investigated using atomistic techniques such as molecular dynamics (MD). However, including temperature effects in these techniques is accomplished through explicit time integration, resulting in high computational costs. This limits the access to the time scales of interest and leads to unrealistic conditions such as extremely high strain rates. Thus, the majority of the studies are limited to 0K scenarios, leaving the temperature dependence of such properties an open problem with significantly less data than its 0K counterpart. The solution we propose looks at phase-space atomic trajectories in a statistical fashion. This allows us to bypass the explicit time integration of MD, effectively coarse-graining in time in a quasistatic fashion, while dedicating more computational resources to reaching larger length scales. Specifically, in this work we use the Gaussian Phase Packets (GPP) statistical approach \cite{gpp}. This framework assumes a Gaussian distribution function for the positions and momenta of all atoms that can be leveraged to evolve the system towards thermodynamic stable states in an efficient way. In this work, we use the GPP formulation to study thermodynamic and kinetic properties of grain boundaries at finite temperatures. We compute the free energies of symmetric tilt grain boundaries via the GPP framework at low to moderately high temperatures and compare them with MD data. Moreover, we use the GPP approach to study how temperature affects their response to quasistatic loading.