Large-eddy simulations (LES) are one of the leading approaches for the numerical modelling of practically relevant turbulent flows. LES, however, rely on closure models that rely on a range of assumptions, such as equilibrium, downward kinetic energy cascade, etc., which can be violated in complex flows and in extreme regimes. On the other hand, direct numerical simulations (DNS) are typically restricted to simplified geometries and low Reynolds numbers. Additionally, DNS, which do not capture the full system geometry and thus large scales driving the flow, require external forcing to achieve steady state. Such forcing is usually described with a single scalar energy-injection rate parameter, which cannot capture all the large-scale complexities of realistic flows. We describe a recently proposed multi-fidelity method, which extracts a detailed large-scale flow structure from an LES through filtering and introduces this data in another fully resolved simulation set up in a specific LES sub-region. Thus, LES acts as a large-scale model for the companion “small-eddy simulation”, or SES, which is responsible for recovering all the small scales that LES does not resolve. The resulting “Large/Small Eddy Simulation” (L/SES) method combines the fidelity of DNS with the ability of LES to capture realistically complex large scales. This technique is agnostic of the details of the system geometry, free from assumptions of equilibrium, homogeneity and isotropy, and is applicable to any realistic complex flow, both incompressible and compressible. The accuracy of the method is assessed for homogeneous isotropic turbulence using both a priori and a posteriori analysis, and its performance is compared with more traditional mesh refinement techniques. We show that L/SES can achieve accuracy comparable to a conventionally forced DNS, with orders of magnitude reduction in computational cost. Various low- and high-order small-scale statistics obtained using the method are shown to be in agreement with the literature. Finally, implications of the method for reacting flows are discussed.