The empirical potential
$$U(x)=\frac{a}{x^{12}} -\frac{b}{x^6}$$
is known as the Lennard-Jones (or 12-6) potential. It is ubiquitous in theoretical chemistry as an approximation of real interatomic potentials. It's "made up" but captures essential features of interatomic potentials, has been particularly useful because of its mathematical simplicity, as explained in another answer, and continues to be broadly implemented, for instance in MD simulation packages.
Is the potential energy function here an actual one for quantities like bond length, or just made up for the sake of this question?
Note that the real potential is "known" only to the interacting matter. As the number of interacting particles (nuclei, electrons) increases the number and complexity of interactions grows staggeringly fast. We do our best to approximate the interactions to facilitate predictions. Even QM simulations are slow to perform on all but the smallest systems. An L-J potential is a pair-wise mean-field potential and as such brushes under many of the complexities of real interactions. You might also find useful information on uses of the L-J potential in our sister site Matter Modeling.