Motion modeling is a critical component in flow-based Video Frame Interpolation (VFI). Existing paradigms either simply consider linear combinations of bidirectional flows or directly predict bilateral flows with the condition of timestamps, lacking the capability of effectively modeling spatiotemporal dynamics in real-world videos. To address this limitation, in this study, we introduce Generalizable Implicit Motion Modeling (GIMM), a novel and effective approach to motion modeling for VFI. Three key designsenable our GIMM as an effective motion modeling paradigm for VFI. First, to obtain useful motion priors for bilateral flow estimations at given timestamps, we perform normalization over scales and directions for initial bidirectional flows. Second, we design a motion encoding pipeline to extract spatiotemporal motion latent from bidirectional flows, effectively representing input-specific motion priors. Third, we predict arbitrary-timestep optical flows within two adjacent input frames via an adaptive coordinate-based neural network implicitly, with spatiotemporal coordinates and motion latent as inputs. Our GIMM can be smoothly integrated with existing flow-based VFI works without further modifications. We show that GIMM performs better than the current state ofthe art on the VFI benchmarks.Code and models will be released to facilitate future research.