One of the most important features of Tesseract is full training capability. Out of the box, the current version of Tesseract is trained to recognize a few dozen fonts and languages. It allows to be additionally trained for almost any font in almost any language.
That also means that Tesseract can be trained for recognition of almost any predefined set of symbols. This capability proves to be very useful for CustomOCR solutions given the nature of text recognition problems being solved. It is often required to do recognition of uncommon fonts, rare characters sets, or known characters with unusual shapes.