The purpose of this study is to propose predictive machine learning models for estimating the compressive strength of ultra high performance fiber reinforced concrete (UHPFRC). Two machine learning models—namely, the random forest model (RFM) and the artificial neural network model (ANNM)—were developed using a dataset of 600 experimental test results with seventeen input parameters. The prediction results showed that both RFM and ANNM demonstrated high accuracy and reliability in estimating the compressive strength of UHPFRC for both train and test datasets. The correlation coefficient values exceeded 0.96 for all models, while the root mean squared error values were below 10% of the average test results. Among the two models, the RFM outperformed the ANNM in foretasting the UHPFRC compressive strength. Sensitivity analysis revealed that curing time, contributing to up to 42%, was the most influential parameter impacting the UHPFRC compressive strength, whereas blast furnace slag, contributing to up to 0.8%, had the least impact.