The field of geospatial research is moving towards the development of more sophisticated multimodal models that can effectively integrate and analyze different types of data, such as overhead remote sensing data, ground-level imagery, and geographic information. This direction is driven by the need to improve the accuracy and generalizability of geospatial models across various tasks, spatial scales, and temporal contexts. Notable papers in this area include: GAIR, which proposes a novel multimodal geo-foundation model architecture that integrates overhead RS data, street view imagery, and their geolocation metadata. LocDiffusion, which leverages diffusion as a mechanism for image geolocalization and achieves competitive performance on benchmark tasks. HiRes-FusedMIM, which introduces a pre-trained model that leverages high-resolution RGB and DSM data for building-level remote sensing applications and demonstrates state-of-the-art performance on several downstream tasks.