Amazon Bedrock adds in-region Claude Opus 5 and Sonnet 5 in Seoul and Singapore

Amazon Bedrock now supports in-region inference for Anthropic's Claude Opus 5 and Claude Sonnet 5 in Seoul, and for Claude Sonnet 5 in Singapore. According to AWS, requests are processed within the Region that is called and do not leave it.
- Claude Opus 5 and Claude Sonnet 5 are available in Seoul, and Claude Sonnet 5 in Singapore, with in-region inference on Amazon Bedrock.
- According to AWS, requests are processed within the called Region and do not leave it.
- Throughput is bounded by the Region's capacity and per-Region service quotas, with no cross-Region routing layer.
- Billing follows standard on-demand pricing for the Region called.
- Access is via the bedrock-runtime endpoint with the Messages, InvokeModel and Converse APIs.
What is available and where
AWS says Amazon Bedrock now supports Claude Opus 5 and Claude Sonnet 5 in the Asia Pacific (Seoul) Region (ap-northeast-2), and Claude Sonnet 5 in the Asia Pacific (Singapore) Region (ap-southeast-1). Both are offered with in-region inference on the bedrock-runtime endpoint.
AWS points to organisations with local data processing requirements in South Korea or Singapore, for example in financial services, healthcare and the public sector. These customers can now use the Anthropic models at scale, according to the company.
For the most current information on model availability in each Region, AWS refers readers to the Regional availability by models page in the Amazon Bedrock User Guide.
How in-region inference works
With in-region inference, a request is processed entirely within the single AWS Region the customer specifies. Input prompts and output results stay in that Region for the full lifecycle of the request, AWS says.
This differs from cross-Region inference profiles. AWS states that there is no routing layer, so a request sent to Seoul or Singapore is served by that Region alone. The trade-off is capacity: throughput is bounded by the Region's capacity, and requests are subject to per-Region service quotas.
Billing follows standard on-demand pricing for the Region being called. Quota consumption, Amazon CloudWatch metrics and AWS CloudTrail log entries are all scoped to that same Region, so there is no source-versus-destination distinction to account for in monitoring.
Getting started
AWS recommends the bedrock-runtime endpoint for new applications. Models are invoked with direct model IDs such as anthropic.claude-opus-5 or anthropic.claude-sonnet-5. The endpoint supports Anthropic's Messages API as well as the Amazon Bedrock InvokeModel and Converse APIs. It also works with Amazon Bedrock features such as Amazon Bedrock Guardrails and intelligent prompt routing.
Developers can try the models without writing code in the text playground of the Amazon Bedrock console. The steps described are:
- Open the console in the Region to use.
- Under Test, choose Playground and select a model.
- Search for anthropic.claude-opus-5, select On-Demand under Inference, and choose Apply.
- Enter a prompt and choose Run.
For programmatic access, AWS provides examples using the AWS SDK for Python (Boto3) with the InvokeModel and Converse APIs, and the Anthropic SDK with the Messages API. The prerequisites are an active AWS account with Amazon Bedrock access, the AWS CLI, Python 3.8 or later, Boto3, the Anthropic SDK and the Amazon Bedrock Token Generator. Usage, performance and costs can be monitored through CloudWatch and AWS Cost Explorer.
What it means for customers
For teams in South Korea and Singapore with strict data residency rules, the launch offers a way to run Claude models while keeping inference inside the chosen Region. The limits are also clear from the announcement: Opus 5 is available only in Seoul, and capacity is capped by per-Region quotas, since there is no cross-Region routing to absorb extra demand.
