OpenAI GPT OSS and NVIDIA Nemotron Models Available on Amazon Bedrock in AWS GovCloud (US)

Amazon Web Services has expanded its Bedrock AI service to include OpenAI's open-weight GPT OSS models and NVIDIA's Nemotron models within AWS GovCloud (US) regions. The new offerings include OpenAI's GPT OSS models in 120B and 20B parameter configurations, alongside NVIDIA's Nemotron variants ranging from Nano 9B v2 to Super 120B models. These additions provide government and regulated industry customers access to both small and large language models through Bedrock's unified API, allowing developers to select optimal models for specific use cases without modifying application code. The models are powered by Mantle, AWS's new distributed inference engine designed specifically for large-scale machine learning model serving on Amazon Bedrock. Mantle introduces enhanced serverless inference capabilities with advanced quality of service controls, automated capacity management, and native compatibility with OpenAI API specifications. Both the OpenAI GPT OSS and NVIDIA Nemotron models feature open-weight architectures with publicly available datasets and training recipes, addressing enterprise requirements for transparency and auditability in AI deployments. The availability in AWS GovCloud (US) specifically targets government agencies and organizations requiring enhanced security and compliance controls for AI workloads. This deployment maintains AWS's enterprise-grade security features while providing the scaling and cost-optimization capabilities necessary for large-scale generative AI implementations in regulated environments.