We present a new library of semi-empirical stellar population models that are based on the empirical MILES and semi-empirical sMILES stellar libraries. The models span a large range of age and metallicity, in addition to an [α/Fe] coverage from -0.2 to +0.6 dex, at MILES resolution (FWHM = $2.5\,$ Å) and wavelength coverage ($3540.5-7409.6\,$ Å). These models are aimed at exploring abundance ratios in the integrated light from stellar populations in star clusters and galaxies. Our approach is to build SSPs from semi-empirical stars at particular [α/Fe] values, thus producing new SSPs at a range of [α/Fe] values from sub-solar to super-solar. We compare these new SSPs with previously published and well-used models and find similar abundance pattern predictions, but with some differences in age indicators. We illustrate a potential application of our new SSPs, by fitting them to the high signal-to-noise data of stacked SDSS galaxy spectra. Age, metallicity, and [α/Fe] trends were measured for galaxy stacks with different stellar velocity dispersions and show systematic changes, in agreement with previous analyses of subsets of those data. These new SSPs are made publicly available.