Claude Platform on AWS supports all AWS commercial regions. That is a deliberately broad answer, and it contrasts with Amazon Bedrock, which maintains its own list of regions where each Claude model is available. Because Claude Platform on AWS is Anthropic-operated — AWS provides the authentication layer, IAM access control, and Marketplace billing, while the models run on Anthropic-managed infrastructure — the platform does not need model capacity physically deployed per AWS region the way a regional managed service does. Since region lists evolve, confirm current availability in the official documentation or on the Claude Platform on AWS service page in the AWS Console before committing an architecture.
What the region actually determines
Your region choice fixes three things. First, the endpoint: requests go to https://aws-external-anthropic.{region}.api.aws/v1/..., and the SigV4 signature must be computed for that same region — a mismatch on either side yields a generic signature-rejection error rather than a helpful diagnostic. Second, the workspace: every workspace is bound to a single AWS region at creation (it binds to the region of the endpoint you called to create it), and a workspace created in us-west-2 can only be accessed through the us-west-2 endpoint. Third, AWS-side governance: the workspace ARN (arn:aws:aws-external-anthropic:{region}:{account-id}:workspace/wrkspc_...) carries the region, and IAM scoping, CloudTrail events, and billing are scoped there. If you use customer-managed encryption keys, the AWS KMS key must live in the same AWS region as its attached workspace.
What the region does not determine
Here is the part that surprises teams coming from Bedrock: the workspace's AWS region does not pin where model inference runs. Inference executes on Anthropic-managed infrastructure, may route to Anthropic's primary cloud, and the data may not reside in AWS at all. If you have data-location requirements, the tool for that is a different, explicit control: the inference_geo parameter.
inference_geo | Behavior | Pricing |
|---|---|---|
"global" (default) | Inference may run in any Anthropic-operated data center | Standard |
"us" | Inference runs only in US data centers | 1.1x multiplier on all token categories |
inference_geo is supported on Claude Opus 4.6, Claude Sonnet 4.6, and later models; older models such as Opus 4.5, Sonnet 4.5, and Haiku 4.5 return a 400 error if the parameter is included. You can set it per request, or govern it at the workspace level with default_inference_geo (applied when a request omits the parameter) and allowed_inference_geos (requests outside the list fail). The response's usage object reports where inference actually ran. Notably, this parameter is available on the first-party Claude API and Claude Platform on AWS, but not on Bedrock or Vertex AI, where region is instead set by the endpoint or inference profile you call.
Choosing a region in practice
Latency: put the endpoint region near your application, since that is where your requests enter the platform — co-locating with your existing AWS workloads is the natural default. Data location: do not reason from the region name; reason from inference_geo and from the platform's data-processing model (Anthropic is the data processor for inference content). If your requirement is that data stay within the AWS environment with AWS as sole processor, the official guidance is to use Claude in Amazon Bedrock instead. Multi-region setups: because workspaces are region-bound, a presence in two regions means two workspaces, each addressed through its own endpoint — plan IAM policies and configuration (AWS_REGION, ANTHROPIC_AWS_WORKSPACE_ID) accordingly. Capacity-wise, the platform runs on a separate pool from both the first-party API and Bedrock, so multi-platform failover is also an option.
inference_geo for residency; use Bedrock when the requirement is "data never leaves AWS."Where to go next
Workspace mechanics are covered in understanding the AWS workspace ID and multi-workspace strategy. If region behavior is pushing you toward Bedrock, read the decision guide before deciding.