Lifestyle

AWS: Claude on Amazon Bedrock can now run entirely inside Seoul and Singapore

AWS says Amazon Bedrock now offers in-region inference for Claude Opus 5 and Claude Sonnet 5 in Seoul and for Claude Sonnet 5 in Singapore. Requests and data are processed only in the AWS Region the customer calls, which is aimed at organizations such as financial services, healthcare and the public sector that must keep data processing local.

About 5 min read

AWS: Claude on Amazon Bedrock can now run entirely inside Seoul and Singapore
Image: Mokaair (Original editorial artwork)

What AWS announced

According to an announcement on AWS's Artificial Intelligence blog, Amazon Bedrock now offers two Claude models made by Anthropic. Claude Opus 5 and Claude Sonnet 5 are available in the Asia Pacific (Seoul) Region (ap-northeast-2), and Claude Sonnet 5 is available in the Asia Pacific (Singapore) Region (ap-southeast-1). An AWS "Region" is a geographic location where AWS runs its data centres. Both offerings run as "in-region inference" through the bedrock-runtime endpoint, which is the web address that applications send their requests to. "Inference" is the step where an AI model processes a request and produces a response.

AWS said the feature is designed for users with local data processing requirements in South Korea or Singapore, such as financial services, healthcare and the public sector. These are often called data residency requirements: rules or policies saying that data must be handled within a particular place. According to AWS, these users can now use the Anthropic models at scale.

AWS: Claude on Amazon Bedrock can now run entirely inside Seoul and Singapore
Mokaair editorial verification flow · Image: Mokaair (Original editorial artwork)
Read the full description

Sources are collected, independently checked, then reviewed by Jev.

Availability in the two Regions, per the AWS announcement
ItemSeoulSingapore
Region codeap-northeast-2ap-southeast-1
Supported Claude modelsClaude Opus 5, Claude Sonnet 5Claude Sonnet 5
Endpointbedrock-runtimebedrock-runtime
Data processing scope (as stated by AWS)Seoul Region onlySingapore Region only

What "in-region inference" means

AWS explained that with in-region inference, requests are processed entirely within the single AWS Region the user specifies. This is different from cross-Region inference profiles, which can pass requests between Regions. In-region inference has no such routing layer: the prompts a user sends and the answers the model returns stay in that Region for the whole life of the request.

AWS also described the trade-offs. Throughput, meaning how many requests can be handled, is limited by that Region's capacity. Requests are also subject to per-Region service quotas, which are usage limits set for each Region. Billing follows standard on-demand pricing for the Region called. Quota use, CloudWatch metrics and CloudTrail logs, which are AWS's monitoring and activity-record tools, are all kept to the same Region. This means monitoring does not have to separate a source Region from a destination Region.

Who this affects

For the general public, infrastructure changes like this usually do not show up directly in the apps people use every day. AWS said the feature is meant to help organizations with local data processing requirements in South Korea or Singapore, such as financial services, healthcare and the public sector, meet data residency needs.

For developers, AWS said the bedrock-runtime endpoint can be called with direct model IDs, such as anthropic.claude-opus-5 or anthropic.claude-sonnet-5. It supports the Anthropic Messages API as well as Amazon Bedrock's InvokeModel and Converse APIs, which are different ways for software to send requests to the models. It also works with features such as Amazon Bedrock Guardrails and intelligent prompt routing. AWS recommends this endpoint for new applications. AWS also said the models can be tried without writing any code in the text Playground of the Amazon Bedrock console.

Limitations and practical notes

  • For the Singapore Region, the announcement lists only Claude Sonnet 5, not Claude Opus 5.
  • According to AWS, throughput for in-region inference is limited by one Region's capacity and quotas, so high-traffic applications need to check whether that is enough.
  • AWS said usage, performance and cost can be monitored through CloudWatch and AWS Cost Explorer to help applications scale as demand grows.

Frequently asked questions

Which Regions and models are covered?

According to the AWS announcement, the Seoul Region (ap-northeast-2) offers Claude Opus 5 and Claude Sonnet 5. The Singapore Region (ap-southeast-1) offers Claude Sonnet 5. All of these use in-region inference.

How does in-region inference differ from cross-Region inference?

AWS said in-region inference has no routing layer between Regions. Requests are processed only by the Region called, and data does not leave that Region. The trade-off is that throughput is limited by that Region's capacity and quotas.

Which organizations are most likely to benefit?

AWS named users with local data processing requirements in South Korea or Singapore, such as financial services, healthcare and the public sector.

Has the pricing model changed?

AWS said billing follows standard on-demand pricing for the Region called. The announcement did not give specific prices.

Where do these details come from?

All of the information comes from AWS's own announcement on its official blog.

Browse the latest news in this topic

More on this topic

Latest travel guides

Sources

Lifestyle