<p>Short food supply chains (SFSCs) have emerged as a sustainable alternative to conventional distribution systems, emphasizing direct-to-consumer delivery or distribution through at most one intermediary. This paper addresses the operational and logistical challenges of SFSCs in a robust optimization framework. We develop three distinct mixed-integer linear programming (MILP) formulations with the objective of minimizing the total system cost—comprising production, operation, transportation, and shortage costs—while adhering to strict regulatory delivery constraints. This offers a comparative analysis of their structural characteristics and theoretical performance. To solve large-scale scenarios where exact solvers often fail, we propose a hybrid genetic-simulated annealing (GSA) framework where a fitness evaluation technique bridges mathematical programming and metaheuristics. Furthermore, an <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\varepsilon \)</EquationSource> <EquationSource Format="MATHML"><math> <mi>ε</mi> </math></EquationSource> </InlineEquation>-greedy method is introduced to adaptively balance exploration and exploitation within the selection and mutation operators. Numerical experiments on ten synthetic datasets ranging from small to very large (up to 11 million variables and 21 million constraints) confirm the robustness of our approach. Results show that the hybrid GSA algorithm consistently delivers high-quality solutions with an average optimality gap of 2.4%, outperforming traditional exact methods in computational stability and time efficiency for large-scale logistics planning. A case study on the Seine-Maritime SFSC network in Normandy, France—comprising 133 producers, 15 intermediaries, and 249 consumers across two food types and a heterogeneous fleet of four vehicle types—demonstrates the practical utility of the proposed framework. Sensitivity analyses further quantify the impact of key parameters on network structure and cost efficiency.</p>