Mobile edge computing (MEC) has been considered as a promising technique to support computation-intensive and latency-sensitive multimedia applications (e.g., video analytics and real-time target tracking) in future 5G/6G systems, by enabling mobile terminals to offload computation workloads to edge servers. In this paper, we study a computation offloading scenario with hybrid multiple access methods to minimize the offloading latency. Particularly, we present a hybrid Non-orthogonal Multiple Access (NOMA)-Orthogonal Multiple Access (OMA) transmission, which covers three offloading strategies, namely, hybrid NOMA, pure NOMA and pure OMA. Based on the successive convex approximation technique, we then design an efficient algorithm to solve the complicated non-convex problem and rigorously prove its convergence. Extensive numerical tests show that our scheme can always help mobile terminals to flexibly choose the best offloading strategy according to mobile terminals' power and energy budgets.