Spatial artificial intelligence is driving a quiet revolution in inference-based landscape modeling. In landscape architecture, established forms of digital representations, based on discrete data structures and geometries, favor stable surfaces and bounded objects at the expense of the atmospheric, temporal, and porous qualities through which landscapes are encountered. Radiance fields, including representations structured through 3D Gaussian splatting (3DGS) and Neural Radiance Fields (NeRF), encode landscapes as continuous volumetric fields of light, generating navigable 3D environments from image-based observations.This paper positions radiance fields as a distinct and emerging category of landscape representations and examines their relevance for landscape architecture through SHARP (Single-image High-Accuracy Real-time Parallax), a novel model by Apple that synthesizes metric-scale 3D Gaussian radiance fields from a single image, preserving absolute spatial dimensions relevant to bodily experience and design reasoning. Trained on large-scale synthetic and real-world data, SHARP infers spatial structure, scale, and visibility from minimal input, collapsing fragmented sensing and modeling processes into a unified inference framework.Infrared imagery captured in Dublin (Ireland), including at the National Botanic Gardens and Killiney Hill Park, serves as input to generate a series of 3D Gaussian reconstructions. In this context, SHARP enables high-fidelity, near-instantaneous synthetic landscape modeling, providing an accessible framework for design research and education. In addition to our model explorations, the approach is further validated through pedagogical experiments with students in a landscape architecture studio at UCD Dublin.The paper frames radiance fields as an emerging form of spatial artificial intelligence and employs SHARP to substantiate a shift from technically fragmented and epistemically stratified processes toward quasi-intuitive, inference-based engagements with landscapes. Rather than definitive spatial records, radiance fields serve as probabilistic spatial propositions whose validity lies in coherence and situated inference rather than exhaustive measurement and uniform sampling, reframing landscapes as fields of conditions shaped by an internalized set of relationships. Recognizing radiance fields as an operative paradigm for sensing and modeling landscapes clarifies their role as a distinct representational framework for landscape architecture.