Networks of evolutionary processors (NEPs) are distributed word rewriting systems typically viewed as language generators. Each node contains a set of words, a set of operations (typically insertion, deletion or rewriting of one symbol with another one), an input filter and an output filter. The purpose of this paper is to overview existing models of NEPs, their variants and developments. In particular, besides the basic model, hybrid networks of evolutionary processors (HNEPs) have been extensively studied. In HNEPs, operations application might be restricted to specific end of the string, but the filters are random-context conditions (they were regular in the basic model). We will also cover the literature on the so-called obligatory HNEPs, i.e., ones where the operations are obligatory: the string that cannot be rewritten is not preserved. Some specific aspects that we pay attention to are: computational universality and completeness, the topology of the underlying graph, the number of nodes, the power of filters.