In this work, we present our MCNP6.1 modeling of some critical fast experimental benchmarks, aiming to qualify our cross section libraries derived from ENDF/B-VII.1, ENDF/B-VII, JEFF-3.1, JENDL-3.3, and JENDL-4.0. The analyzed benchmarks are characterized by simple geometries which helps to have taken precise results, and concerning the type: HEU-MET-FAST (highly enriched uranium). Those benchmarks are extracted from the International Handbook of Evaluated Criticality Safety Benchmark Experiments (IHECSBE) published by the Nuclear Energy Agency [1]. A detailed comparison of the results of our simulation was made in order to highlight the influence of these nuclear data types on our calculations, due to its importance for the stability of nuclear reactors. We interpreted the difference between calculation and experiment ($$C{-}E$$) for the principal parameter $$k_{\textrm{eff}}$$ through a comparison of the fission and capture rates of the major fissile elements. The different spectral indices F28/F25, F49/F25, F37/F25, C28/F25, and F23/F25 at the cores center are also calculated. For the majority of the studied HEU (highly enriched uranium) benchmark cases, The ENDF/B-VII and JEFF-3.1 have a good agreement with the experimental ones, concerning the $$k_{\textrm{eff}}$$ results. The average discrepancy from the experimental values for ENDF/B-VII is 0.42$$\%$$, and 0.39$$\%$$ for JEFF-3.1. An overestimation was observed for most evaluations concerning benchmarks with tungsten carbonate reflectors. The best results were obtained by JENDL-3.3, with a maximal discrepancy $$C{-}E$$ estimated 0.47$$\%$$ concerning fission rate, and 4.25$$\%$$ for capture rate. In analyzing the spectral indices, for GODIVA and FLATTOP-25, the best results were obtained by JENDL-4.0 with a maximal discrepancy of 2.66$$\%$$.