{"id":"nvidia-kumo","slug":"nvidia-kumo","name":"NVIDIA Kumo","orgId":"nvidia","family":"nvidia-kumo","releaseDate":null,"availability":"announced","openWeights":true,"license":null,"paramsTotal":null,"paramsSource":null,"contextWindow":null,"knowledgeCutoff":null,"modelCardUrl":"https://x.com/NVIDIAAI/status/2112847569827348804","hfRepo":null,"modalities":["tabular","time-series"],"aliases":["NVIDIA Kumo","nvidia kumo","kumo structured","nvidia kumo structured","kumo tabular","kumo-tabular","nvidia/kumo-tabular","kumo relational","kumo-relational","nvidia/kumo-relational","kumo forecast","kumo-forecast","nvidia/kumo-forecast","kumo anomaly","kumo-anomaly","nvidia/kumo-anomaly"],"aliasNote":null,"origin":"tracked","tracked":{"trackedSince":"2026-10-21","announced":"2026-10-21","blurb":"NVIDIA's family of open foundation models for structured business data, presented at GTC Berlin as Kumo Structured, part of the NVIDIA Agent Knowledge Foundation. Four models have weights on Hugging Face: Kumo Tabular (single-table classification and regression), Kumo Relational (predictions across related tables), Kumo Forecast (time-series forecasting) and Kumo Anomaly (time-series anomaly detection). Tabular and Relational predict in-context, without task-specific training. Not LLMs.","api":"none","openWeights":"yes","params":{"text":"Kumo Tabular 28M / 62M / 215M (Small / Medium / Large); Kumo Forecast ~340M frozen MOMENT-1-large encoder plus a trained head; Kumo Anomaly 8M; Kumo Relational not stated","source":"NVIDIA Hugging Face blog and model cards"},"facts":[{"label":"Models with weights","value":"nvidia/Kumo-Tabular, nvidia/Kumo-Relational, nvidia/Kumo-Forecast and nvidia/Kumo-Anomaly on Hugging Face"},{"label":"Licenses","value":"Kumo Tabular and Kumo Relational weights: OpenMDW 1.1. Kumo Forecast and Kumo Anomaly: Apache 2.0"},{"label":"Kumo Tabular","value":"In-context tabular Transformer (column, row and in-context attention), pretrained only on synthetic tables from structural causal models; up to 10 classes per forward pass. Weights and blog published Sep 29, 2026"},{"label":"Kumo Relational","value":"Based on KumoRFM-2; fine-tunes TabICLv2 weights (BSD-3-Clause) and adds relational message passing; also served as a Kumo Relational NIM"},{"label":"Kumo Forecast / Anomaly","value":"Run with the Kumo-TS library. Forecast pairs a MOMENT encoder with retrieval of historical patterns (DARR); Anomaly is diffusion-based with adaptive thresholds"},{"label":"Libraries","value":"structured-data-models (Apache 2.0, GitHub) for Tabular and Relational; Kumo Engine, an open library stack for building and deploying structured-data models"},{"label":"Snowflake","value":"NVIDIA says Snowflake customers can use Kumo Engine and Kumo Relational in Snowflake Notebooks; NVIDIA's docs include a Snowflake Notebooks guide"},{"label":"Kumo AI","value":"Kumo Relational is based on KumoRFM from Kumo AI. Fortune and The Information reported in June 2026 that NVIDIA had acquired Kumo AI; neither company formally announced it","unsure":true,"note":"press"},{"label":"\"Kumo Time Series\"","value":"Named in one NVIDIA blog post; no model card or weights under that name. NVIDIA's docs list Forecast and Anomaly as the time-series models (run via Kumo-TS)","unsure":true}],"claims":["Lab claim: Kumo Tabular ranks first on TabArena (Elo 1950), BeyondArena, TALENT and ScoringBench, and runs 17x faster than LimiX-2 on a single RTX 6000 Pro.","Lab claim: Cognite, Coinbase, Databricks, DoorDash, Revolut and Tubi are among the first to work with Kumo models; Databricks pairs Kumo Anomaly with Databricks Genie Ontology."],"sources":[{"kind":"lab","title":"NVIDIA AI announcement on X","url":"https://x.com/NVIDIAAI/status/2112847569827348804","date":"2026-10-21"},{"kind":"lab","title":"NVIDIA Open Models Advance Europe's Sovereign Intelligence Across Industries (NVIDIA blog)","url":"https://blogs.nvidia.com/blog/open-models-europe-industry/","date":"2026-10-21"},{"kind":"lab","title":"NVIDIA Agent Toolkit Expands, Grounding Enterprise Agents in Company Data (NVIDIA blog)","url":"https://blogs.nvidia.com/blog/enterprise-platforms-agent-toolkit/","date":"2026-10-21"},{"kind":"lab","title":"NVIDIA Kumo overview (NVIDIA docs)","url":"https://docs.nvidia.com/sdgm/about-nvidia-kumo/overview","date":"2026-10-21"},{"kind":"lab","title":"NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier (Hugging Face blog)","url":"https://huggingface.co/blog/nvidia/kumo-tabular","date":"2026-09-29"},{"kind":"repo","title":"Kumo Tabular on Hugging Face","url":"https://huggingface.co/nvidia/Kumo-Tabular","date":"2026-10-21"},{"kind":"repo","title":"Kumo Relational on Hugging Face","url":"https://huggingface.co/nvidia/Kumo-Relational","date":"2026-10-21"},{"kind":"repo","title":"Kumo Forecast on Hugging Face","url":"https://huggingface.co/nvidia/Kumo-Forecast","date":"2026-10-21"},{"kind":"repo","title":"Kumo Anomaly on Hugging Face","url":"https://huggingface.co/nvidia/Kumo-Anomaly","date":"2026-10-21"},{"kind":"press","title":"Exclusive: Nvidia snaps up Kumo AI in latest acquisition (Fortune)","url":"https://fortune.com/2026/06/03/nvidia-snaps-up-kumo-ai-in-latest-acquisition/","date":"2026-06-03"}],"firstBenchmark":null},"organization":{"id":"nvidia","name":"NVIDIA","slug":"nvidia","country":null,"url":null,"color":"#76B900","mark":"N"},"price":null,"answer":{"word":"Pending benchmarks.","scope":"Announced Oct 21, 2026; no independent results yet.","lead":"As of Oct 22, 2026, NVIDIA Kumo is pending benchmarks on isitsota: announced Oct 21, 2026, with no independent results on the boards we track yet.","confidence":"low"},"status":"PENDING_BENCHMARKS","verdicts":[]}