JAX-Fluids is a computational fluid dynamics solver that leverages JAX (a machine learning framework from Google) to enable automatic differentiation through numerical simulations. Written entirely in JAX, the solver supports CPU/GPU/TPU execution and scales across HPC clusters. It solves Navier-Stokes equations using finite-volume methods with high-order schemes and provides automatic differentiation capabilities for end-to-end optimization of physics-based numerical models.