Ctenophores (comb jellies) are the largest animals in the world who locomote primarily using cilia. Despite the relative simplicity of this propulsive system, they possess unusually impressive maneuvering capabilities. Ctenophores group their cilia in coordinated platelike bundles called ctenes, which are arranged in eight rows circumscribing the body. Ctenes in each row are metachronally coordinated, but each row’s frequency and beat direction can be independently controlled. This coordination allows the animals to swim forward and backward at nearly the same speed, and to turn rapidly with small turning radii. This surprising agility appears across a wide range of animal sizes, and bridges the gap between the viscous-dominated (low Reynolds number) and inertial-dominated (high Reynolds number) fluid dynamic regimes. To quantify the locomotion of freely swimming ctenophores, we used deep learning-based kinematic tracking to reconstruct animal trajectories, extracting the position and orientation of the animal during complex three-dimensional maneuvers. These measurements allow us to calculate performance parameters such as (e.g.) minimum length- specific turning radii, maximum angular and translational velocities and accelerations, and backward to forward swimming speed ratio (B:F). Our results show that at similar angular velocities ctenophores can achieve a length- specific turning radius two times smaller than other centimeter scale zooplankton. Furthermore, they also exhibit the unusual case of a B:F close to one. By quantifying the swimming behavior of ctenophores, we provide a first step toward the potential development of bio-inspired devices, sensors, and vehicles that may be able to leverage similar systems in the intermediate Reynolds number regime.