Kumo Relational NIM Container Overview
Description
Kumo Relational NIM v1.1 is the latest release. It provides classification and regression inference over connected, multi-table data. It packages the Kumo Relational model, its inference runtime, and dependencies in a downloadable container. Clients provide labeled context rows, related tables, declared relationships, and rows to score; the model produces predictions without task-specific training or weight updates.
The earlier v1.0 release remains available under different model terms and supports additional task types. The license and permitted uses below depend on the container version selected.
For v1.1, the container components are ready for commercial or non-commercial use. The v1.0 model is licensed for non-commercial research or evaluation purposes only.
License/Terms of Use
GOVERNING TERMS: All versions of the NIM container are governed by the NVIDIA Software License Agreement and the Product-Specific Terms for NVIDIA AI Products. Model terms differ by version:
| Container version | Model terms and use |
|---|---|
| v1.0 series | The NVIDIA License applies to the model. Use is limited to non-commercial research or evaluation. |
| v1.1 | The OpenMDW License Agreement, version 1.1 applies to the model. Commercial and non-commercial use are permitted subject to its terms. The model is derived from TabICLv2 pretrained weights; their BSD 3-Clause License terms, copyright notice, and disclaimer also apply. |
The v1.1 model terms do not change the terms applicable to v1.0 images. Check the license materials included with the version you deploy.
You are responsible for ensuring that your use of NVIDIA provided models complies with all applicable laws.
Deployment Geography
Global.
Release Date
- v1.0 NGC container: 08/18/2026 via Kumo Relational on NGC.
- v1.1 NGC container: October 2026, version 1.1.0 via Kumo Relational on NGC; image repository
nvcr.io/nim/nvidia/kumo-relational. - v1.1 model weights on Hugging Face: 09/23/2026 via nvidia/Kumo-Relational.
Program Classes
The container includes the following model:
| Model Name & Link | Use Case | How to Pull the Model |
|---|---|---|
| Kumo Relational | Classification and regression over relational data without task-specific training. | Automatic: model artifacts are packaged in the container; no runtime model download is required. |
The v1.0 profile also supports forecasting and temporal link prediction. These tasks are outside the v1.1 NIM profile; its supported tasks are classification and regression.
Deployment Details
Pull the Kumo Relational image from nvcr.io/nim/nvidia/kumo-relational using NGC credentials and select the tag for the version you intend to use. Run it on a supported NVIDIA GPU. The inference API supports one-shot predictions and reusable sessions. See the Kumo Relational NGC container page for available image tags and deployment instructions. Use the license and task rules for the selected version above.
Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.
Reference(s)
- Kumo Relational NGC container page
- Kumo Relational on NVIDIA Build (API playground and deployment guidance)
- Kumo Relational model weights
- Structured Data Models source code
- Kumo Relational API reference
- v1.0 Kumo Relational SDK getting started
Container Version(s)
- v1.0 series — earlier release; model use limited to non-commercial research or evaluation under the NVIDIA License.
- 1.1.0 — latest release; model use governed by OpenMDW-1.1, with TabICLv2 BSD 3-Clause terms also applicable.
Security Common Vulnerabilities and Exposures (CVEs)
Review the Security Scanning tab on NGC for the latest scan results. NVIDIA provides Vulnerability Exploitability eXchange (VEX) documents for certain open-source vulnerabilities in the same tab.
Ethical Considerations
NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. Developers should work with their internal developer team to ensure these software components meet requirements for the relevant industry and use case and address unforeseen product misuse.
Please report quality, risk, security vulnerabilities or NVIDIA AI Concerns here.
Get Help
Getting started with the NIM
Deploying and integrating the NIM is straightforward thanks to our industry standard APIs. Visit the NVIDIA NIM documentation for deployment and integration guidance.
NVIDIA Developer Community Forum
Get access to community knowledge and support cases in the NVIDIA Developer Forums.