In the first part of my talk, I will review the phenomenology required to alleviate the Hubble tension through late-time new physics, focusing on the so-called angular (2D) and anisotropic (3D) BAO data, which themselves are in mutual tension. I will then present a recent model, the wXCDM, which combines quintessence with an exotic component known as “phantom matter.” This component satisfies the strong energy condition but exhibits negative energy density and positive pressure. The wXCDM model outperforms its competitors and resolves the H₀ tension when tested against a comprehensive dataset that includes 2D BAO. However, when angular BAO is replaced by 3D BAO — as expected — the model can no longer yield high values of H₀. Nevertheless, it still produces low chi-squared values, comparable to those found with the CPL parametrization. Finally, I will describe the Weighted Function Regression method, which enables a Bayesian and model-agnostic reconstruction of the effective dark energy properties and the late-time cosmic expansion history. I will also assess the impact of supernova data on quantifying the statistical evidence for dynamical dark energy in the late universe.