The adaptive fixed-time bipartite containment issue of p-normal multiagent systems (MASs) with unmodeled dynamics is investigated. In view of the structural characteristics of high-order systems, the technology of adding a power integrator is adopted which overcomes the difficulty caused by the high power of virtual control and real control items. In addition, to further optimize the rationality of resource utilization, an improved threshold strategy including saturation threshold, control input and decreasing function of errors is proposed. Based on practical fixed-time theory, a fixed-time containment control strategy is devised, which ensures that some followers converge to a convex hull spanned by leaders, others converge to the symmetric one spanned by leaders within fixed time. Finally, the effectiveness of the developed containment control scheme and the superiority of the improved threshold strategy are illustrated by comparison simulation results.