One proposed component of the upcoming Deep Underground Neutrino Experiment (DUNE) near detector complex is a multi-purpose, magnetized, gaseous argon time projection chamber: the Multi-Purpose Detector (MPD). We explore the new-physics potential of the MPD, focusing on scenarios in which the MPD is significantly more sensitive to new physics than a liquid argon detector, specifically searches for semi-long-lived particles that are produced in/near the beam target and decay in the MPD. The specific physics possibilities studied are searches for dark vector bosons mixing kinetically with the Standard Model hypercharge group, leptophilic vector bosons, dark scalars mixing with the Standard Model Higgs boson, and heavy neutral leptons that mix with the Standard Model neutrinos. We demonstrate that the MPD can extend existing bounds in most of these scenarios. We illustrate how the ability of the MPD to measure the momentum and charge of the final state particles leads to these bounds. arXiv:1912.07622v1 [hep-ph] 16 Dec 2019 Liquid Argon Near Detector: The liquid argon near detector has a width (transverse to the beam direction) of 7 m, a height of 3 m, and a length (in the beam direction) of 5 m. The fiducial mass of the near detector is 50 t of argon. See Ref. [2] for more detail.Multi-Purpose Detector: The Multi-Purpose Detector (MPD) is a magnetic spectrometer with a cylindrical high-pressure gaseous argon time projection chamber (HPTPC) at its heart. The HPTPC has a diameter of 5 m and a length of 5 m. It is surrounded by an electromagnetic calorimeter (ECAL) and the HPTPC+ECAL system is situated inside a magnet with 0.5 T central field. The axis of the cylinder is perpendicular to the beam direction. However, for simplicity, when simulating our signals (see Section 3), * Some projections of DUNE operation include the possibility of an upgraded beam (roughly twice the number of protons on target per year) that can operate in the second half of the experiment. We assume a constant number of protons on target per year, so our ten-year projection is equivalent to a shorter operation time with such an upgraded beam. † This is in contrast with the LArTPC, where the conversion distance of photons is O(10 cm). Far more photons can fake the signal of e + e − in the LArTPC than in the HPTPC. * Ref. [63] provides a thorough review of searches for dark photons and existing constraints. * The results of this analysis in antineutrino mode are qualitatively equivalent.