The current stage of development of the world economy is characterized by a gradual increase in the level of its intellectualization and the transition to an economy based on knowledge rather than material resources. The ability of the national economy to absorb knowledge and use it effectively will increasingly determine the economic strength of a nation and the well-being of its citizens. The key components of a knowledge-oriented economy are innovations and organizations that contribute to their creation and dissemination. Achieving and maintaining an advantage over competitors is facilitated not only by innovation and education, but also by forming effective relationships between enterprises, which provides the conditions for the creation of network structures - clusters. Diffusion model is one of the methods of the distribution process innovation research. The subject of the diffusion model is the idea of the level of prevalence of innovation among a certain set of potential consumers in terms of mathematical dependence on the time that has elapsed since the introduction of innovation. Today, the most well-known and popular theories of innovation are the diffusion models of Frank Bass and Everett Rogers. From the point of view of the agent-oriented model of diffusion, the buyers of innovation cluster products are autonomous agents who are guided by certain rules and differ not so much in their fixed assets and implemented technologies as in production and innovation strategies. Depending on the behavior of agents, the following types are selected: innovator agent, «early minority» agent, «early majority» agent, skeptic agent, conservative agent. The general scheme of perception of innovation in agent-oriented diffusion model of innovation clusters has a hierarchical form. The higher the level of knowledge and competence of the agent, the greater his propensity to innovate. Innovators are the first agents to embrace or acquire innovation. As a result, we got a model that shows the number of customers at the time of model time. The developed agent model of diffusion in innovation clusters can be used in marketing research and product pricing, taking into account consumer behavior according to the classification of E. Rogers.