The Imperial Valley hosts a network of active strike‐slip faults that comprise the southern San Andreas fault (SAF) and San Jacinto fault systems and together accommodate the majority of relative Pacific–North American plate motion in southern California. To understand how these faults partition slip, we model the long‐term mechanics of four alternative fault networks with different degrees of connectivity through the Imperial Valley using faults from the Southern California Earthquake Center Community Fault Model version 5.0 (v.5.0). We evaluate model results against average fault‐slip rates from the Uniform California Earthquake Rupture Model v.3 (UCERF3) and geologic slip‐rate estimates from specific locations. The model results support continuous linkage from the SAF through the Brawley seismic zone to the Imperial and to the Cerro Prieto faults. Connected faults decrease surface strain rates throughout the region and match more slip‐rate data. Only one model reproduces the UCERF3 rate on the Imperial fault, reaching the lower bound of . None of the tested models reproduces the UCERF3 preferred rate of . In addition, high‐strain energy density rates around the Cerro Prieto fault in all models suggest that the UCERF3 preferred rate of may require revision. The Elmore Ranch fault‐slip rate matches the UCERF3 rate only in models with continuous linkage. No long‐term slip‐rate data are available for the El Centro and Dixieland faults, but all models return less than on the El Centro fault and on the Dixieland fault. This suggests that the Dixieland fault may accommodate a significant portion of plate‐boundary motion.