Proteasome inhibitors are promising compounds for a number of therapies, including cardiovascular and eye diseases, diabetes, and cancers. We previously reported a series of furan-based peptidic inhibitors with moderate potencies against the proteasome β5 subunit, hypothesizing that the C-terminal furyl ketone motif could form a covalent bond with the catalytic residue, threonine 1. In this context, we describe further optimizations of the furan-based peptides, and a series of dipeptidic and tripeptidic inhibitors were designed and synthesized, aiming at improved potency and better solubility. Most of the tripeptidic inhibitors demonstrated improved potency and selectivity as β5 subunit inhibitors in both enzymatic and cellular assays, and good antineoplastic activities in various tumor cell lines were also observed. However, no inhibitory effects were observed for the dipeptidic compounds, which led us to presume that a noncovalent binding mode is adopted. Docking studies and molecular dynamics simulations were carried out to verify this presumption, with results showing that the distance between the furyl ketone motif and Thr1 is slightly too long to form covalent bond.