A new global coastal tide model for non-polar regions is presented: GOAT-1C. The GOAT-1C atlas makes use of an admittance-based empirical tide estimation approach to infer residual coastal tidal variability omitted by the FES2022 tidal atlas from a combination of OpenADB nadir altimetry and, for the first time, wide-swath observations from the Surface Water and Ocean Topography mission. This approach builds on the Global Operator-based Admittance Tidal process framework (GOAT) which treats the global oceanic tidal response as a continuous function that can be learned using a DeepONet neural operator. The model shows improved performance in coastal regions in terms of variance reduction, tide gauges, and shoreline imagery relative to FES2022. Independent SWOT Cal/Val validation indicates that approximately 1 % of coastal sea-level variance attributable to tidal-model error is associated with spatial structure resolved only at kilometre scales. Furthermore, using an ensemble approach, we provide per-constituent uncertainty estimates which allow users to understand where the product should be used and how much to trust it. Inclusion of wide-swath observations improves fine-scale corrections but must be handled with care. The final model and uncertainty estimates are released on a 4 km coastal grid.