Run NVIDIA Nemotron and OpenAI GPT OSS models on Amazon Bedrock in AWS GovCloud (US)

This article announces the availability of OpenAI GPT OSS and NVIDIA Nemotron open-weight models on Amazon Bedrock in AWS GovCloud (US), enabling secure AI inference for government agencies.

• OpenAI GPT OSS models (120B and 20B) and NVIDIA Nemotron models (Nano 9B v2, Nano 12B v2, Nano 30B, Super 120B) now available in AWS GovCloud (US)
• Inference runs entirely within AWS GovCloud (US) boundary on U.S.-operated infrastructure, meeting FedRAMP High, DoD SRG, ITAR, and CJIS compliance requirements
• Zero operator access design prevents AWS, customer, or model provider access to inference data like prompts and completions
• Supports multiple endpoints: bedrock-mantle (OpenAI-compatible API) and bedrock-runtime (AWS SDK with native Bedrock features)
• In-Region inference available in us-gov-west-1; Geo cross-Region inference routes across us-gov-west-1 and us-gov-east-1 while staying within GovCloud boundary
• Standard, Priority, and Flex service tiers supported; Reserved tier not currently available
• Enables agentic applications for security assessments, document analysis, contract review, and compliance checking

Government agencies can now build and scale generative AI applications with advanced open-weight models while maintaining data residency and compliance within AWS GovCloud (US).