Whether case should be ignored, or not, depends on the application. Whether the model really ignores it or not, depends on how the model has been trained. The fastText models that I use are case-sensitive. However, if I use them in the application that should be case-insensitive, I convert each word to lower case before calculating its embedding.