The finite-time cluster synchronization (FTCS) problem in multi-weighted fractional-order coupled two-layer neural networks (MFOCTLNNs) is investigated in this paper. It is worth noting that the considered model consists of two-layer networks with distinct topologies and dynamic behaviors, and there exist multiple connections between nodes within each layer. First, by applying fractional-order finite-time convergence lemmas and creating a state-feedback control scheme, we establish a FTCS criterion for the drive network. Subsequently, we compress the nodes in the same cluster of the drive network into a virtual node. In order to achieve the FTCS of MFOCTLNNs, we then design another new delayed controller to synchronize the response network with these virtual nodes. Finally, we provide an illustrative example to validate the effectiveness of the proposed results.