Journal of Information Organization Hajar Ait Lamkademe 16 2 2026 https://doi.org/10.6025/jio/2026/16/2/45-58 https://www.dline.info/jio/fulltext/v16n2/jiov16n2_1.pdf The rapid proliferation of Large Language Models (LLMs) and multimodal AI systems has created an urgent need for systematic, analyzable resources to track architectural innovation, organizational strategy, and computational scaling trends. This paper presents a structured analysis of the "LLMs & Frontier AI Models Dataset" (Kaggle, 2026), a curated tabular resource documenting metadata for 47 attributes across notable AI models released through mid-2026. We examine organizational productivity patterns, frontier model compute allocations, strategic divergences between closed and open weight development paradigms, and the shifting balance between industrial and academic contributions. Our findings reveal a highly concentrated ecosystem dominated by a small cohort of well-resourced laboratories, with distinct strategic specializations emerging across geographic and organizational boundaries. We discuss implications for research reproducibility, policy development, and future dataset curation practices, while acknowledging limitations inherent to static, crowd-sourced metadata collections.