LaTeXML supports the standard LaTeX mechanism for handling non-ASCII encodings of the input TeX sources: using the inputenc package. The LaTeXML binding of inputenc loads the encoding definition (generally with extension def) directly from the LaTeX distribution (which are generally well-enough behaved to be easily processed). These encoding definitions make the upper 128 code points (of 8 bit) active and define TeX macros to handle them.
Using the commandline option --inputencoding=utf8 to latexml allows processing of sources encoded as utf8, without any special packages loaded. [future work will make LaTeXML compatible with xetex]