Amazon Bedrock in AWS GovCloud Adds NVIDIA Nemotron and OpenAI GPT OSS Models

In July 2026, Amazon Web Services expanded Amazon Bedrock in the AWS GovCloud (US) cloud, adding support for OpenAI's open GPT OSS models and the NVIDIA Nemotron series. It is the first time these models have run on sovereign infrastructure for the US public sector.

Which models are now available in GovCloud?

AWS GovCloud users now have access to two families of open frontier models: OpenAI's open-weight GPT versions and the full NVIDIA Nemotron lineup.

• OpenAI GPT OSS 120B — the flagship open model, with 120 billion parameters
• OpenAI GPT OSS 20B — a lighter version for resource-constrained workloads
• NVIDIA Nemotron Nano 9B v2 — a compact second-generation model for edge scenarios
• NVIDIA Nemotron Nano 12B v2 — an expanded Nano with improved reasoning capabilities
• NVIDIA Nemotron Nano 30B — a mid-size model balancing answer quality and inference speed
• NVIDIA Nemotron Super 120B — the series flagship, for complex analytical tasks

All the models are open-weight: their parameters are published, so the architecture and behavior of the system can be verified independently. This is a critical requirement for government customers, who must audit AI systems before deploying them in sensitive scenarios.

Why government agencies need GovCloud

AWS GovCloud (US) is an isolated cloud region for US federal, state and defense organizations. It meets FedRAMP High, DoD Impact Level 2/4 and ITAR requirements: data physically stays in the US, access is limited to US citizens, and the entire stack has been audited for compliance with government security standards.

Before this announcement, government organizations on Bedrock worked mainly with proprietary models such as Anthropic Claude and Amazon Titan. Open-weight models change the picture fundamentally: agencies get systems with transparent weights that are suitable for independent verification. That reduces the "black box" risk and lets agencies run their own safety evaluations before deployment.

Data residency inside GovCloud ensures that queries to the model and its responses do not leave the protected region. This matters most when handling citizens' personal data, intelligence material and national security documentation.

How to get started with the new models

Amazon Bedrock offers several inference modes in GovCloud: through the API using the AWS SDK, through the Bedrock web console, and with support for managed inference providers for flexible load distribution. Two service tiers are available: on-demand (pay per call) and Provisioned Throughput (reserved capacity for predictable performance).

NVIDIA Nemotron models are trained with a focus on enterprise and government tasks, including the use of RLHF on data reflecting corporate and government policies. The lineup covers a wide range, from lightweight Nano versions for fast inference to the powerful Super 120B for complex analysis.

What this means

The arrival of frontier open-weight models in AWS GovCloud marks a new milestone for government AI: transparent, auditable models with sovereign data storage are becoming the standard for regulated sectors. The competition between NVIDIA and OpenAI for a share of the government segment is now playing out directly on federal-level infrastructure.