The field of geospatial modeling and analysis is experiencing a significant shift towards the integration of large language models (LLMs) and vision-enabled agents. This new direction is enabling the automation of complex environmental modeling tasks, such as hydrologic modeling and geospatial data integration, and is lowering the barrier between Earth observation data, physics-based tools, and decision makers. The use of LLMs is also improving the accuracy and equity of geospatial predictions by incorporating geostatistical priors and spatially structured reasoning. Furthermore, the development of agentic workflows and fine-tuned small language models is increasing the efficiency and scalability of geospatial tasks. Notable papers in this area include AQUAH, which introduces an end-to-end language-based agent for hydrologic modeling, and GeoSR, which proposes a self-refining agentic reasoning framework for geospatial knowledge boundaries. Additionally, GeoFlow presents a method for automatic generation of agentic workflows for geospatial tasks, and Fine-Tuning Small Language Models for Autonomous Web-based Geographical Information Systems demonstrates the feasibility of browser-executable models for AWebGIS solutions.