AWS says Claude Opus 5.5 and Sonnet 5.5 are available through Bedrock in GovCloud, where organizations can configure Claude Code for regulated development.
AWS announced on October 5, 2026, that Claude Opus 5.5 and Claude Sonnet 5.5 are available through Amazon Bedrock in AWS GovCloud (US). The company’s technical guide explains how organizations can connect those models to Claude Code for software development in environments with regulatory or compliance requirements, including workloads involving ITAR.
The guide distinguishes the models’ authorization status: Opus 5.5 and Sonnet 5.5 have FedRAMP Class D certification, while Claude Sonnet 5—not Sonnet 5.5—has DoD Impact Level 4 and 5 authorization as well as FedRAMP Class D certification. AWS cautions that organizations must assess the approach against their own compliance obligations; the availability of a model in GovCloud does not by itself establish that a particular workload meets every applicable requirement.
AWS describes two ways to access the models. The bedrock-runtime endpoint is available in both GovCloud regions, US-East and US-West, and supports Bedrock Guardrails and invocation logging, features AWS recommends for applications that need audit trails. The bedrock-mantle endpoint, available in GovCloud US-West, supports Anthropic’s Messages API and features including server-side tools, background inference, and Projects. Guardrails and invocation logging are exclusive to bedrock-runtime.
With Claude Code, developers can use the models from a terminal, supported integrated development environments such as VS Code and JetBrains, or background workflows. The tool can edit code across files, run commands and tests, work with Git history, and connect to systems such as Terraform and Kubernetes through the Model Context Protocol.
AWS’s deployment guidance also emphasizes identity controls, model-version pinning, usage limits, and security review. It notes that inference runs through Bedrock, but Claude Code operates on developers’ local machines and should be assessed separately under an organization’s security policies.