I/O-efficient algorithms take the advantage of large capacities of external memories to verify huge state spaces even on a single machine with low-capacity RAM. On the other hand, parallel algorithms are used to accelerate the computation and their usage may significantly increase the amount of available RAM memory if clusters of computers are involved. Since both the large amount of memory and high speed computation are desired in verification of large-scale industrial systems, extending I/O-efficient model checking to work over a network of computers can bring substantial benefits. In this paper we propose an explicit state cluster-based I/O efficient LTL model checking algorithm that is capable to verify systems with approximately $10^{10}$ states within hours.