PyMC now supports jax and numba as well, and can use the numpyro samplers. https://www.pymc.io/projects/examples/en/latest/samplers/fas...
Between all these tools we hope the modern bayesian can find the one that works best for them