Context. One of the main challenges in observational cosmology is related to the redshift evolution of the average hydrogen (HI) ionization in the Universe, as evidenced by the changing in ionization level of the intergalactic medium (IGM) through cosmic time. Starting from the first cosmic reionization, the rapid evolution of the IGM physical properties in particular poses severe constraints for the identification of the sources responsible for maintaining its high level of ionization up to lower redshifts. Aims. In order to probe the ionization level of the IGM and the ionization capabilities of bright quasi-stellar objects (QSOs) at z = 4, we selected a sample of 2508 QSOs drawn from the Sloan Digital Sky Survey (SDSS, DR14) in the redshift interval 3.6 ≤ z ≤ 4.6 and absolute magnitude range −29.0 ≲ M₁₄₅₀ ≲ −26.0. Particularly, we focus on the estimate of the escape fraction of HI-ionizing photons and their mean free path (MFP), which are fundamental for characterizing the surrounding IGM. Methods. Starting from UV/optical rest-frame spectra of the whole QSO sample from the SDSS survey, we estimated the escape fraction and free path individually for each of the QSOs. We calculated the Lyman continuum (LyC) escape fraction as the flux ratio blueward (∼900 Å rest frame) and redward (∼930 Å rest frame) of the Lyman limit. We then obtained the probability distribution function (PDF) of the individual free paths of the QSOs in the sample and studied its evolution in luminosity and redshift, comparing our results with those in literature. Results. We find a lower limit to the mean LyC escape fraction of 0.49, in agreement with the values obtained for both brighter and fainter sources at the same redshift. We show that the free paths of ionizing photons are characterized by a skewed distribution function that peaks at low values, with an average of ∼49 − 59 proper Mpc at z ∼ 4, after possible associated absorbers (AAs) were excluded. This value is higher than the one obtained at the same redshift by many authors in the literature using different techniques. Moreover, the PDF of free path gives information that is complementary to the MFP derived through the stacking technique. Finally, we also find that the redshift evolution of this parameter might be milder than previously thought. Conclusions. Our new determination of the MFP at z ∼ 4 implies that previous estimates of the HI photoionization rate Γ_HI available in the literature should be corrected by a factor of 1.2−1.7. These results have important implications when they are extrapolated at the epoch of reionization.