The accurate prediction of end-point carbon content is a key technology for realizing the intelligentization of the converter steelmakingConverter steelmaking processProcess. However, the practical application of mechanism modelMechanism model (MM) is often limited by the difficulty of determining their hyperparameters, while purely data-driven models generally lack interpretability. To overcome these limitations, this study proposes a hybrid model (HM) that integrates mechanism-based and data-driven approaches. Specifically, unmeasurable parameters in the mechanism modelMechanism model are optimized using a genetic algorithm, after which the mechanism modelMechanism model is coupled with a random forestRandom Forest (RF) to construct the hybrid framework. The results demonstrate that the hybrid model substantially outperforms both stand-alone mechanism-based and purely data-driven models. Within the error interval of (− 0.006, 0.006), the prediction hit rate reached 78.5%, and within (− 0.01, 0.01), it reached 94.0%. Compared with alternative models, the hybrid approach achieved the best performance, with a mean absolute error (MAE) of 3.97% and a root mean square error (RMSE) of 5.19%, thereby confirming its effectiveness.