The State of Solvers for Energy Planning - Results of the 2025 Open Energy Benchmark

This report presents the 2025 results of the Open Energy Benchmark initiative, a transparent and reproducible benchmarking effort evaluating optimization solvers on realistic, community-contributed energy system models. The benchmark compares four open-source solvers (CBC, GLPK, HiGHS, and SCIP) and a proprietary baseline (Gurobi) across a diverse set of LP and MILP problems originating from 13 modelling frameworks. Using 213 benchmark problems, collected in `instances.tar`, the report evaluates solver performance in terms of runtime, success rate, and scalability under standardized computational settings. The analysis highlights how solver capabilities vary depending on problem class, model formulation, and problem scale, while also exploring the evolution of solver performance over time and the computational challenges associated with increasingly large and complex energy planning models. The purpose of this report is not to identify a single "best" solver or produce a universal ranking of solver packages. Solver performance depends strongly on the characteristics of the optimization problem, modelling framework, configuration choices, and user requirements. Instead, the benchmark is intended as an educational and diagnostic resource that helps energy modelers better understand solver behavior, helps solver developers identify opportunities for improvement, and helps funding organizations identify areas where investments in open-source optimization software can have the greatest impact.