NVIDIA introduces Kumo Relational for database-native predictions
Discovery source: NVIDIA on Hugging Face · Source date:
NVIDIA says Kumo Relational is a pre-trained relational foundation model that makes predictions directly on enterprise databases without task-specific training, feature engineering, or table flattening. In NVIDIA’s Oct. 21, 2026 announcement, the model is described as using historical context examples in a single forward pass to predict outcomes such as churn, demand, or fraud, and as being built to work for both humans and agents through NVIDIA’s structured-data-models and kumo-engine libraries or as an NVIDIA NIM. NVIDIA also says its paper reports better results than supervised baselines on 41 tasks across 15 databases, while noting limitations on tasks with very large training sets and on signals that are too many hops away.
Check the sources
This is a NextGen-AI briefing, not a reproduction of the discovery source. A newsletter may cover several developments; consult the evidence below for this briefing’s specific subject. The source date refers to the discovery entry, not necessarily the announcement date.