Scores how well the answers of a language model to closed survey questions represent a human population. The Population Fidelity Score combines accuracy, adaptability and structure (da Silva et al., 2026) <doi:10.48550/arXiv.2609.36253> and is reported for the pooled population, for every demographic subgroup and for paired comparisons of a tuned and a base model. The package also computes the distances, alignment scores, dispersion, structure, reliability and uncertainty measures used to compare synthetic and human survey samples, each catalogued with its origin and its application. A 'Rust' core shared with the 'Python' package gives both languages the same numbers.