NVIDIA released the 30-billion-parameter Nemotron 3 Large Telco Model on its blog, saying it delivers accuracy gains for telecom-specific tasks. It is the company's first major update to its Nemotron family of open models since the release of the Nemotron 3 model.
NVIDIA reported the Nemotron 3 Large Telco Model achieved a benchmark score of 89% in the Artificial Analysis Intelligence Index v4.3.2, measured on reasoning, coding, scientific and agentic workloads. That compares with the earlier benchmark score of 78% for the Nemotron 3 model.
The Nemotron 3 Large Telco Model is built on open weights, training data and recipes and targets telecom-specific workflows such as network configuration and customer incident triage. Availability begins with the release of the full fine-tuning recipe, initially for telecom operators.
"Open models allow SoftBank Corp. to build on the rapid progress of global foundation models while applying the network knowledge and operational expertise we have accumulated over many years," said Rajeev Koodli, principal fellow of SoftBank Corp. and senior vice president of SB Telecom America. SoftBank is using the model in developing its SoftBank Large Telecom Model.
The announcement follows NVIDIA's collaboration with partners to turn open models into practical building blocks for telecom AI. NVIDIA said the model is a critical step in enabling operators to deliver locally adapted AI services.
NVIDIA did not say how the model will perform in real-world deployments, and raised the open question of how operators will balance model customization with data privacy. The company said it will release tools to help operators fine-tune the model with their own operational data.
Source: nvidia
