TLDRocket
Sign in

Introducing Anthropic models on Amazon Bedrock for in-region inference in Seoul and Singapore

Amazon Web Services Aamna Najmi ● Covered by 2 sources

Anthropic’s Claude Opus 5 and Sonnet 5 are now in AWS Bedrock in Seoul, plus Sonnet 5 in Singapore. AWS says the data stays in-region the whole time, which is the real hook.

Based on reporting by Amazon Web Services, Aamna Najmi — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Amazon Bedrock has added Anthropic’s Claude Opus 5 and Claude Sonnet 5 for in-region inference in Seoul, and Claude Sonnet 5 in Singapore. The models run on the bedrock-runtime endpoint, so customers with data residency rules in places like financial services, healthcare, and the public sector can keep inference inside the Region they choose.

That is the main point here, and AWS is being unusually explicit about the boundary. There’s no routing layer. A request sent to ap-northeast-2 or ap-southeast-1 is handled there alone, and both the input prompt and the output stay in that Region for the life of the request. The tradeoff is capacity: throughput is limited by that Region, and requests are subject to per-Region service quotas.

Billing follows standard on-demand pricing for the Region you call. CloudWatch metrics, CloudTrail logs, and quota usage are also scoped to that same Region, which simplifies monitoring because there’s no source-versus-destination split to track.

For new applications, AWS recommends the bedrock-runtime endpoint. You can call it with direct model IDs like anthropic.claude-opus-5 or anthropic.claude-sonnet-5, and it supports Anthropic’s Messages API as well as Bedrock’s InvokeModel and Converse APIs. AWS also says Bedrock features such as Guardrails and intelligent prompt routing work here.

The post walks through setup in the Bedrock console and with code. It shows examples using Boto3 in Singapore, the Converse API in Seoul, and Anthropic’s Messages API through the anthropic SDK with the aws_bedrock_token_generator. AWS also points users to CloudWatch and Cost Explorer for watching usage, performance, and costs as demand grows.

My take — AI-written commentary, not fact-checked reporting

This is the right kind of cloud announcement: boring in the best way. Data residency is not a marketing flourish, it is the whole product for plenty of buyers, and AWS is leaning into that instead of pretending every model belongs everywhere. The closed-model crowd will call it control; the rest will call it compliance with a working API.

Read more about this at: Amazon Web Services

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.