Compare AIFind AIAI NewsAI How-To
About Us
PrivacyTermsFAQContactContact
AIB Inc.Company info
© 2026 AIB Inc.

AWS Adds Claude Models for Seoul and Singapore Inference

AWS Adds Claude Models for Seoul and Singapore Inference

AWS ML Blog·Thursday, October 1, 2026
  • •AWS added Claude Opus 5 and Claude Sonnet 5 in Seoul, plus Claude Sonnet 5 in Singapore, for in-region inference.
  • •Requests, prompts and results stay in the selected Seoul or Singapore Region throughout processing.
  • •Regional capacity and quotas limit throughput; APIs, console access and monitoring tools are supported.
  • •AWS added Claude Opus 5 and Claude Sonnet 5 in Seoul, plus Claude Sonnet 5 in Singapore, for in-region inference.
  • •Requests, prompts and results stay in the selected Seoul or Singapore Region throughout processing.
  • •Regional capacity and quotas limit throughput; APIs, console access and monitoring tools are supported.
  • •AWS added Claude Opus 5 and Claude Sonnet 5 in Seoul, plus Claude Sonnet 5 in Singapore, for in-region inference.
  • •Requests, prompts and results stay in the selected Seoul or Singapore Region throughout processing.
  • •Regional capacity and quotas limit throughput; APIs, console access and monitoring tools are supported.
  • •AWS added Claude Opus 5 and Claude Sonnet 5 in Seoul, plus Claude Sonnet 5 in Singapore, for in-region inference.
  • •Requests, prompts and results stay in the selected Seoul or Singapore Region throughout processing.
  • •Regional capacity and quotas limit throughput; APIs, console access and monitoring tools are supported.

Amazon Web Services added Anthropic’s Claude Opus 5 and Claude Sonnet 5 to Amazon Bedrock for in-region inference in Seoul, and Claude Sonnet 5 in Singapore, on September 30, 2026. Requests sent through the bedrock-runtime endpoint are processed entirely in the AWS Region selected by the customer; prompts and generated results stay there for the full request lifecycle. AWS cites local data-processing requirements in South Korea and Singapore, including in financial services, healthcare and the public sector, as use cases.

In-region inference sends each request only to the Seoul Region (ap-northeast-2) or Singapore Region (ap-southeast-1), without the routing layer used by cross-Region inference profiles. The trade-off is that throughput is limited by the selected Region’s capacity, and requests are subject to per-Region service quotas. On-demand billing follows the standard price for the Region called. Quota use, Amazon CloudWatch metrics and AWS CloudTrail log entries are also scoped to that Region, with no source-versus-destination distinction in monitoring.

AWS says new applications can call the service using direct model IDs such as anthropic.claude-opus-5 and anthropic.claude-sonnet-5. Supported interfaces include Anthropic’s Messages API and Amazon Bedrock’s InvokeModel and Converse APIs; the service also supports Amazon Bedrock Guardrails and intelligent prompt routing. Customers can try the models in the Bedrock console text playground by selecting a model, entering a prompt and running it, without coding or SDK setup.

For programmatic use, AWS lists an active account with Bedrock access, AWS CLI, Python 3.8+, Boto3, Anthropic SDK and the Amazon Bedrock Token Generator as prerequisites. The post describes examples using the AWS SDK for Python with InvokeModel, Converse API, and Anthropic Messages API through the Anthropic SDK. AWS says customers can monitor usage, performance and costs with CloudWatch and AWS Cost Explorer. The launch makes the listed Claude models available for applications requiring inference to remain within the selected Seoul or Singapore Region.

Amazon Web Services added Anthropic’s Claude Opus 5 and Claude Sonnet 5 to Amazon Bedrock for in-region inference in Seoul, and Claude Sonnet 5 in Singapore, on September 30, 2026. Requests sent through the bedrock-runtime endpoint are processed entirely in the AWS Region selected by the customer; prompts and generated results stay there for the full request lifecycle. AWS cites local data-processing requirements in South Korea and Singapore, including in financial services, healthcare and the public sector, as use cases.

In-region inference sends each request only to the Seoul Region (ap-northeast-2) or Singapore Region (ap-southeast-1), without the routing layer used by cross-Region inference profiles. The trade-off is that throughput is limited by the selected Region’s capacity, and requests are subject to per-Region service quotas. On-demand billing follows the standard price for the Region called. Quota use, Amazon CloudWatch metrics and AWS CloudTrail log entries are also scoped to that Region, with no source-versus-destination distinction in monitoring.

AWS says new applications can call the service using direct model IDs such as anthropic.claude-opus-5 and anthropic.claude-sonnet-5. Supported interfaces include Anthropic’s Messages API and Amazon Bedrock’s InvokeModel and Converse APIs; the service also supports Amazon Bedrock Guardrails and intelligent prompt routing. Customers can try the models in the Bedrock console text playground by selecting a model, entering a prompt and running it, without coding or SDK setup.

For programmatic use, AWS lists an active account with Bedrock access, AWS CLI, Python 3.8+, Boto3, Anthropic SDK and the Amazon Bedrock Token Generator as prerequisites. The post describes examples using the AWS SDK for Python with InvokeModel, Converse API, and Anthropic Messages API through the Anthropic SDK. AWS says customers can monitor usage, performance and costs with CloudWatch and AWS Cost Explorer. The launch makes the listed Claude models available for applications requiring inference to remain within the selected Seoul or Singapore Region.

Read original (English)·Sep 30, 2026
Infra#amazon bedrock#anthropic#claude opus 5#claude sonnet 5#in region inference#data residency#seoul#singapore#bedrock runtime