AWS Bedrock Inference Profiles Explorer

About This Tool

This tool helps you understand AWS Bedrock inference profile routing - which regions your requests can be processed in when using cross-region inference profiles.

Data Source

Profile data is fetched from the AWS Bedrock ListInferenceProfiles API across 17 AWS regions. We query for SYSTEM_DEFINED profiles and extract:

Source regions: Where you can call the profile from
Destination regions: Where your request may be routed for processing

Update Frequency

Profile data is automatically refreshed daily at 6:00 AM UTC via an AWS Lambda function. After each refresh, the CloudFront cache is invalidated to ensure you see the latest data.

Latency Testing Methodology

When latency test results are shown for a profile, here's how they were measured:

Metric: Time to First Token (TTFT) using the streaming API
Method: A Lambda function in us-west-2 creates Bedrock clients for each source region and invokes the model
Prompt: Minimal test prompt ("Say 'ok' and nothing else") to minimize cost
Samples: Multiple invocations per region to calculate averages

Limitations

We cannot determine the actual destination region where a request was processed without CloudWatch logging enabled
Latency measurements include network round-trip from the Lambda to the Bedrock endpoint
Test results are point-in-time snapshots and may vary based on load and network conditions
Not all profiles have latency test data - tests are run manually for selected profiles

Resources

AWS Bedrock Cross-Region Inference Documentation

Understanding Region Filtering

Profile Types vs Geographic Boundaries

AWS Bedrock inference profiles have a region prefix in their ID (e.g., us., eu., apac.) that indicates their routing scope. This filter lets you find profiles two ways:

Profile Types (e.g., "US profiles")

Filters by the profile's AWS-assigned prefix. For example:

US profiles shows profiles like us.anthropic.claude-*
EU profiles shows profiles like eu.anthropic.claude-*
Australia profiles shows profiles like au.anthropic.claude-*

These profiles route requests only within their designated region scope.

Geographic Boundaries (e.g., "Europe destinations")

Filters by checking that all destination regions fall within a geographic boundary. Use this for data residency compliance:

Europe (all EU destinations) - Shows profiles where every destination region is in Europe
North America (US + Canada destinations) - Shows profiles routing only to US/Canada regions
Asia Pacific (all APAC destinations) - Shows profiles routing only to APAC regions

Which Should I Use?

For data residency compliance: Use geographic boundary filters to ensure inference stays within specific regions
For finding specific profiles: Use profile type filters to quickly find profiles by their AWS naming