Dynamical systems are commonly used to represent real-world processes. Model reduction techniques are among the core tools for studying dynamical systems models, they allow to reduce the study of a model to a simpler one. In this poster, we present an algorithm for computing exact nonlinear reductions, that is, a set of new rational function macro-variables which satisfy a self-consistent ODE system with the dynamics defined by algebraic functions. We report reductions found by the algorithm in models from the literature.