So have you tried compiling GMP from source for your platform? If there’s no lib yet from someone else, you’ll have to do it yourself. https://gmplib.org/ has all the sources you need. Note that there’s also mini-gmp.c and mini-gmp.h there, which is a minimalistic implementation of a subset of GMP functions. If you want the full thing you have to copy and configure the subcomponents (multi-precision integer, arbitrary-precision floats etc.) yourself.
Depending on what exactly you want to do exactly (we don’t know) it might also be feasible to use mbedtls’s bignum library functions, which is readibly available for that platform. Or other libraries such as PlatformIO Registry.