<p>This study presents a lightweight brittle metamaterial beam built from a new hybrid unit cell that merges double-honeycomb and a re-entrant auxetic cell. Stereolithography is used to create the beams. Consequently, finite element simulations are used to validate the brittle behavior of the beam with the experiment. The design of experiments (DoE) uses cell geometry parameters to examine the effects of uniform wall thickness, functionally graded thickness (FGT), and re-entrant angle on bending stiffness, strength, and specific energy absorption (SEA). The results show that the low re-entrant angles result in a stretch-dominated response, with high bending stiffness, strength, and progressive failure. Higher angles promote bending-dominated mechanisms. FGT further improves performance by combining high peak load with extended post‑peak deformation, leading to higher SEA than uniform‑thickness designs. To maximize stiffness, yield strength, and SEA at the same time, a multi-objective optimization framework is developed. The surrogate models are created using neural network-type GMDH polynomials. The multi-objective NSGA-II optimization results in a Pareto front that captures trade-offs to maximize the mechanical performances and SEA. It is revealing that maximum stiffness/strength configurations reduce SEA by 24%, whereas optimized high-SEA designs achieve 246J/kg, significantly exceeding literature values (100-178J/kg) for comparable polymeric beam metamaterials under bending, while maintaining competitive mechanical performances. All things considered, this work establishes a robust surrogate-assisted method for designing brittle metamaterials with high mechanical performances that shows promise for lightweight structures in biomedical implants, automotive crashworthiness, aerospace, and protective gear.</p>