In model predictive control (MPC) applied to high-order systems like LCL filter grid-connected converters - widely used in renewable energy integration - longer prediction horizons are increasingly requisite. However, augmenting the prediction steps leads to an exponential growth in computational time. To address this challenge, this study introduces a variable step prediction horizon tailored for multistep-finite-control-set MPC. This innovative approach permits an extended prediction horizon timespan without incurring in additional computational over-head. Simulation experiments validate the methodology, revealing marked improvements in dynamic performance and steady-state response compared to the traditional approach to this strategy.