Enhancers coordinate gene expression in response to developmental and environmental cues. Plant enhancers, however, lack the readily detectable molecular hallmarks of animal enhancers, hindering large-scale functional characterization. Here, we characterize the species- and condition-specific enhancer activity of over 350,000 sequences derived from accessible chromatin regions of Arabidopsis, tomato, maize, and sorghum. We identify GC content and transcription factor binding sites as key features controlling enhancer strength and demonstrate how they can be used to rationally design synthetic enhancers. Enabled by these data, we developed plantGREP, a deep learning model that predicts enhancer strength and identifies functional sequence motifs. We apply plantGREP to generate constitutive, species-, and condition-specific enhancers, and to locate regions with enhancer activity near important developmental genes in crop genomes. These results facilitate the targeted editing of enhancers in crop genomes and identify strategies for the design of cell-type-specific plant enhancers.