How gene expression programs are coordinated across space and time to build a complete plant remains poorly understood. Here, we combine high-resolution spatial transcriptomics, single-nucleus RNA sequencing, and a gap-free reference genome to build a 3D spatiotemporal atlas of rice, spanning 10 organs and 61 stages from seed to seed. Our datasets profile 851,725 nuclei and 347,640 spatial bins, defining 119 cell types and 133 subtypes across the rice lifecycle. Through multimodal integration, we reconstruct an inferred developmental trajectory graph that links cell states from the proembryo to mature organs and identifies trajectory-associated regulators. We further implicate the pleiotropic regulator OsARF1 in distinct spatial gene programs across developmental contexts. We also uncover widespread axis-associated transcriptional asymmetry, including dorsoventral endosperm programs that partition starch and protein accumulation. Collectively, this high-resolution atlas provides a spatiotemporal framework for understanding rice development and guiding future crop improvement.