In this article is presented and assessed a massive parallel processing model for basic operations with k-mers from genomic sequences, based on defined functions in terms of N-dimensional spaces. The model is implemented using a set of OpenCL cores available at github.com/bioinfud/k-merscl and assessed using a heterogeneous platform CPU/GPU and a dataset based on randomly generated kmers. The results make evident a maximum speed up factor of 4898X and a minimum of 350X.