We propose Permutation Equivariant Graph Neural CDEs, an equivariant and parameter-efficient extension of Graph Neural CDEs for dynamic graph representation learning.