Benefiting from highly tunable pore environments, some metal-organic frameworks (MOFs) have recently shown promising prospects in the separation of methanol-to-olefin (MTO) products (mainly C3H6 and C2H4). However, the "trade-off" between gas storage capacity and selectivity always results in inefficient separation. In addition, poor stability of MOFs also limits practical separation applications. Herein, we have successfully assembled a layered Y-MOF (FJI-W9) with bent diisophthalate ligands (H4L), Y-O chains, and 2-fluorobenzoic acids. As expected, FJI-W9 not only exhibits good chemical stability but also shows significant potential for C3H6/C2H4 separation. For FJI-W9, the C3H6 uptake at 298 K and 10 kPa is 63 cm3/g, and the IAST selectivity of FJI-W9 for C3H6/C2H4 (V/V = 50/50) is calculated to be 20.5. To the best of our knowledge, both C3H6 uptake and selectivity of FJI-W9 surpass most porous materials. GCMC simulation indicates that the special supramolecular binding sites in FJI-W9 have much stronger interactions with C3H6 than C2H4 molecules. More importantly, practical breakthrough experiments demonstrate that FJI-W9 can effectively separate C3H6/C2H4 (50/50) mixtures, thus obtaining high-purity C2H4 and C3H6, respectively.