Fatigue failure, recognized as one of the most prevalent failure modes in engineering structures, remains inadequately understood in terms of its fundamental mechanical mechanisms. Existing fatigue crack growth models are highly dependent on experimental fatigue data while lacking a universal theoretical framework. To overcome these limitations, we develop a Fatigue-Free Calibration Cohesive Zone Model (F-free model), which can efficiently predict fatigue crack growth rates without the need for fatigue data. Through the definition of cohesive endurance limit and its associated separation displacement, a cyclic damage increment triggering criterion is established. The concept of conditional yield stress in elastoplastic materials is extended to the framework of the cohesive zone model. The cohesive endurance limit is determined as the intersection point between the actual traction-separation curve and a straight line parallel to its initial linear segment. The proposed F-free model is validated by comparing its simulated fatigue crack growth rates with experimental data from two key test scenarios: interlaminar delamination in composite laminates and face-core debonding in sandwich structures. The prediction range of this model can effectively encompass the experimental observation results, accurately capturing both the crack growth rates and the Paris' exponent values for mode I interfacial fatigue cracking. The applicability of the F-free model is further evaluated. The fatigue crack growth rates of interlaminar delamination in double cantilever beam (DCB) specimens under different cohesive endurance limits are simulated. The results indicate that the F-free model can provide a prediction region for interfacial fatigue crack growth rates and a prediction range for the Paris' exponent between 0.99 and 6.3. The proposed F-free model is applicable for predicting the fatigue crack growth of elastoplastic materials or at ductile fracture interfaces. This advancement provides a novel theoretical framework for fatigue damage analysis, effectively bridging the gap between empirical observations and mechanical modeling. The proposed F-free model is able to significantly improve the computational efficiency of fatigue damage tolerance analysis.