In this paper we study the non-linear multi-index transportation problem with concave cost functions. We solved the non-linear transportation problem on a network with 5 indices (NTPN5I) described by sources, destinations, intermediate nodes, types of products, and types of transport, that is formulated as a non-linear transportation problem on a network with 3 indices (NTPN3I) described by arcs, types of products, and types of transport. We propose a genetic algorithm for solving the large-scale problems in reasonable amount of time, which was proven by the various tests shown in this paper. The convergence theorem of the algorithm is formulated and proved. The algorithm was implemented in Wolfram Language and tested in Wolfram Mathematica.