# NVIDIA releases telecom model and recipe for operator customization

_Published Tuesday, October 6, 2026 at 10:21 AM EDT · AI, Infrastructure · Latest · Tier 2 — Notable_

![NVIDIA releases telecom model and recipe for operator customization — Primary](https://blogs.nvidia.com/wp-content/uploads/2026/10/image1.png)

NVIDIA announced a 30-billion-parameter Nemotron 3 Large Telco Model and released a recipe for adapting open models to operators' own data. AdaptKey fine-tuned the model on open source telecom datasets for tasks including network configuration and customer incident triage.

The recipe covers the full fine-tuning pipeline using NVIDIA's NeMo libraries, allowing operators to adapt models to their networks, customers and procedures. NVIDIA says production deployment also requires pipelines that anonymize sensitive records and generate privacy-preserving synthetic data, alongside controls for autonomous workflows.

## Sources

- [NVIDIA Blog](https://blogs.nvidia.com/blog/telecom-operators-open-models/)

---
Canonical: https://techandbusiness.org/newswire/p7LIpfF61jspTg7Zd69M-V
Published: 2026-10-06T14:21:41.810Z
Story chronology: 2026-10-06T13:00:19.000Z
Retrieved: 2026-10-06T16:57:43.788Z
Publisher: Tech & Business (techandbusiness.org)