Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- The announcement confirms GPT-5.6 Terra and Luna on bedrock-runtime in AWS GovCloud (US), with the Responses, Converse, and Chat Completions APIs. It defines a specific model, endpoint, and API scope.
- GovCloud (US) Cross-Region Inference automatically routes requests across both GovCloud (US) Regions. AWS describes higher throughput and redundancy without requiring customers to manage capacity across multiple Regions.
- AWS also says Cross-Region Inference for OpenAI models is available in AWS Regions where Bedrock offers those models. That does not mean every Region, model, or account is enabled.
- The announcement says Runtime calls use Bedrock model-invocation logs, CloudWatch metrics, and model-level cost reporting. Account eligibility, exact prices, quotas, and model configuration still need confirmation in AWS documentation and the account.
The direct answer: what this launch covers
In an announcement dated October 30, 2026, AWS said Amazon Bedrock's bedrock-runtime endpoint in AWS GovCloud (US) supports OpenAI's GPT-5.6 models Terra and Luna, with the Responses, Converse, and Chat Completions APIs. The announcement also adds Cross-Region Inference for OpenAI models within AWS GovCloud (US).
The defined scope is Terra and Luna, bedrock-runtime, those three APIs, and cross-region routing within GovCloud (US). The announcement does not say that a particular AWS account is entitled to invoke these models, and it gives no account-specific eligibility rule, exact price, or quota. Treat it as a product announcement, not proof that an account has been enabled.
Confirm the Runtime endpoint and API
This support applies to bedrock-runtime. AWS lists Responses, Converse, and Chat Completions as the available APIs. For an integration, check the current Terra and Luna model cards and calling requirements in the Amazon Bedrock User Guide, linked from the announcement at https://docs.aws.amazon.com/bedrock/.
Do not assume the same API support applies to every other Bedrock product path just because the API names are familiar. The announcement does not list specific model IDs, inference-configuration IDs, or account quotas. Check current model documentation and your AWS account instead of carrying values over from another model or Region.
Separate GovCloud routing from the general Region statement
Within AWS GovCloud (US), Cross-Region Inference automatically routes inference requests across both GovCloud (US) Regions. AWS describes the benefits as higher throughput and increased redundancy, without requiring customers to manage capacity across multiple Regions. AWS says this GovCloud routing maintains US data sovereignty.
The announcement separately says Cross-Region Inference for OpenAI models is available in all AWS Regions where OpenAI models are offered on Amazon Bedrock. That availability is conditional on the models being offered in a Region. It does not say Terra and Luna are available in every AWS Region or confirm availability for any individual account. Read the specific two-Region GovCloud statement separately from the broader availability statement.
Check access in the target account before rollout
Start with the current Terra and Luna model cards in the Amazon Bedrock User Guide, then have an administrator for the target AWS GovCloud (US) account verify that the account can call the required model and API. The announcement does not publish an account-eligibility rule, an activation procedure, or a list of enabled accounts; do not infer that an organization is approved from the launch notice alone.
Confirm that the application actually calls bedrock-runtime, and validate the selected API and routing within the account's AWS permissions and service configuration. A proxy or fixed IP cannot replace AWS account authorization or establish GovCloud access eligibility. Ask AWS through the appropriate account or support channel when account evidence is unclear.
Use existing telemetry to check calls and spend
AWS says the native OpenAI APIs run on bedrock-runtime and use the same account-level controls as other Bedrock models. Model-invocation logs can be delivered to Amazon S3 or Amazon CloudWatch Logs. CloudWatch metrics include invocation counts, token counts, latency, throttles, and errors.
AWS also says usage is itemized by model in Cost Explorer and the AWS Cost and Usage Report. Once access is authorized, use these existing records to verify actual calls and spend. The announcement supplies no exact rate, account-level remaining quota, or evidence that a particular account is enabled, so API support and cross-region routing do not establish price or capacity.
Sources
Frequently Asked Questions
Can an existing Bedrock Managed Agents or bedrock-mantle integration switch directly to Runtime?
Do not infer that from this announcement. It documents three APIs for Terra and Luna on bedrock-runtime, but makes no compatibility or migration promise for Managed Agents or bedrock-mantle. For the separate preview's regions, permissions, and setup scope, see Bedrock Managed Agents public preview: regions, permissions and getting started: https://puppyip.com/en/resources/amazon-bedrock-managed-agents-openai-preview-guide.
Does this Terra/Luna announcement also grant Daybreak Cyber access?
Do not infer that. The announcement names Terra and Luna only; it does not address Daybreak Cyber, its separate access review, or account authorization. For Daybreak Red's regions and access eligibility, see OpenAI Daybreak Red reaches AWS GovCloud: regions and access eligibility: https://puppyip.com/en/resources/openai-daybreak-amazon-bedrock-guide.
What non-sensitive details should I give AWS Support when a call fails?
Provide the API, model name (or ID as shown in your account), Region, failure timestamp, and error code. Do not send access keys, temporary credentials, or other secrets. The announcement does not diagnose account-specific failures; follow AWS's official support process.