It is difficult to keep up with the constant changes in numbers regarding Covid infections and related deaths. This article from The Atlantic does a very good job of explaining how epidemiological models are meant to be used. Briefly, they are not crystal balls that predict the future. They take available data regarding both the characteristics of the disease and current human responses generate a range of possible outcomes. Different models often use different assumptions and sometimes different underlying data resulting in the models returning different results. Additionally, as the underlying data changes, so to does the model. Were a vaccine that is 100% effective be invented tomorrow, the models must be changed drastically. This does not mean the previous model was wrong, it just means it was the best guess we had based on incomplete information.