NVIDIA's KumoRFM-2: a foundation model that predicts churn, fraud and demand straight from relational databases
solyarisoftware · x · 2026-10-23
NVIDIA released Kumo Relational (KumoRFM-2), a small purpose-built foundation model that predicts what happens next — churn, fraud, demand, conversion, risk — directly from your relational database.
- Works across relationships between existing tables instead of flattening everything into one giant table
- No per-task model training or manual feature engineering; supports in-context learning with up to 10,000 historical examples for classification or regression
- Tested on 41 tasks across 15 databases; NVIDIA reports it beating supervised relational deep-learning and expert-engineered gradient-boosted-tree pipelines
- First foundation model shown to surpass supervised approaches on common benchmark tasks, scaling to billion-scale relational datasets
More from Models
- TypeSafe's Jev: a "Decision Model" LLM That Returns Floats at $0.042/M Input Tokens — ariG23498 · 2026-10-23
- Microsoft's MAI-Base-2 scaling ladder pegs the L42 run at ~1.8×10^23 FLOPs — teortaxesTex · 2026-10-23
- Local model agentic coding benchmark: 27B matches 125B, tool-calling is the lever — Joralk_ · 2026-10-23
- NVIDIA open-sources PersonaPlex-7B, a full-duplex voice AI model that listens and talks simultaneously — solyarisoftware · 2026-10-23
- Gemini Analyzes Whether a Simple 'Hi' Is a Jailbreak Attempt in Its Thinking Steps — OldIntroduction2909 · 2026-10-23
- 5 Essential Reads to Master Pi's Agent Harness, from Onboarding to Source Code — solyarisoftware · 2026-10-23