Deep eutectic solvents are mixtures of two compounds characterized by a melting point significantly below the predicted ideal eutectic melting point. Among the most used components for the preparation of deep eutectic solvents is choline chloride, which is produced on the megaton scale and has applications as an animal food supplement. Choline chloride can form eutectic mixtures with a wide range of organic compounds close to room temperature. We studied the capabilities of polarizable force fields to model these systems.[1] Introducing an additional damping function was essential during force field development to screen the charge interactions between the chloride anion and the hydroxyl group of the cation in our developed approach. However, parameters of the non-bonded Thole screening function must be fitted against the first-principles molecular dynamics simulations. Therefore, invariant and equivariant machine learning interatomic potentials were studied next.[2,3] The equivariant Allegro model in combination with an active learning scheme requests solely few thousand DFT calculations of a small system to simulate systems with several thousand atoms on the ns-time scale on a single GPU-node within one day. This facilitates reliable investigation of dynamical properties since at least five simulations are recommended to obtain a well converged average value.[2] Thus, machine learning interatomic potentials provide reliable structural and dynamical properties at a fraction of cost of first-principles molecular dynamics simulations. Our studies on the unique nature of deep eutectic solvents revealed that the incorporation of the chloride anion into the hydrogen bond network of the urea derivative is strongly correlated to the non-ideal mixing behaviour of choline chloride systems.[4] Furthermore, we observe close contacts between two lithium atoms bridged by oxygen atoms of the organic compound in lithium bistriflimide systems.[3] Please note, the close Li-Li-contacts play a minor role in classical force field simulations, even with scaled charges. This highlights limitations of common classical force fields compared to approaches with forces on the accuracy of density functional theory calculations. [1] O. Shayestehpour, S. Zahn, J. Phys. Chem. B, 2022, 126, 3439-3449 [2] O. Shayestehpour, S. Zahn, J. Chem. Theory Comput., 2023, 19, 8732-8742 [3] O. Shayestehpour, S. Zahn, J. Chem. Phys., 2024, 161, 134505 [4] O. Shayestehpour, S. Zahn, J. Phys. Chem. B, 2020, 124, 7586-7597