Theories of planet formation predict the existence of a population of free-floating planets that are not gravitationally tethered to any host star. Gravitational microlensing provides a unique tool for studying these objects. The first results of Sumi et al. (2011) claimed that Jupiter-mass free-floating planets are as common as main-sequence stars. However, these results disagree with censuses of substellar objects in young clusters and star-forming regions and with predictions of planet formation theories. I will present new results of the analysis of a ten times larger sample of microlensing events discovered by the OGLE-IV survey during the years 2010-2018, which shed new light on the population of free-floating planets. I will also discuss prospects for detecting free-floating planets with the future missions, like Euclid and WFIRST.