We introduce a multi-study high-dimensional factor analysis toolbox 'MultiEFM' based on the elliptical factor model framework, which learns latent heterogeneous features and accounts for cross-study variation among sources. It provides robust estimation algorithms for heterogeneous datasets, particularly tailored for multi-study RNA sequencing arrays and complex spatial multi-omics layers. The package implements highly efficient initialization strategies, identifiability constraints alignment, and computationally scalable parameter estimation paradigms.