Stream process mining is a powerful approach to enable analysts to gain real-time business model understanding, however it has certain challenges, primarily related to scalability and performance efficiency, in addition to producing complex and messy models. This paper introduces “Mini-process cubes”, an innovative approach for online process mining that addresses these challenges by decomposing streamed event logs into smaller, memory-efficient submodels (mini-process cubes). Our approach tries to leverage the process cubes that are used in offline contexts as a valuable tool for business analysts, taking into consideration the need to handle efficiency and precision. For this reason, we introduced the Cube Succession Matrix (CSM) and Cube Short Loop Matrix (CSLM) variables that are used as precomputed values to make the mining process faster. We also integrate the offline contribution factor ( \(\gamma \) ) to benefit from the historical data in fixing misordering or the incompleteness of streamed event logs. This research work is designed to reduce model complexity while providing a dynamic analytics tool for business analysts in real-time and bridging the gap between offline and online process mining.