chore(bedrock-alpha): add Claude 4 models and missing Cross region inference profile #35908
Conversation
github-actions
bot
added
p2
distinguished-contributor
| * Anthropic's Claude Haiku 4.5 model, most cost-efficient and fastest. | ||
| * Delivers near-frontier performance with substantially lower cost and faster speeds. | ||
| * | ||
| * Features: | ||
| * - Supports vision (Image input modality) | ||
| * - Cross-region support | ||
| * - Supports Bedrock Agents | ||
| * - Best for: Large-scale deployments, budget-conscious applications, real-time customer service, latency-sensitive use cases | ||
| */ |
There was a problem hiding this comment.
I'm unsure about the extent to which we should describe this section.
I wonder if just mentioning the model name would suffice.
The content is written in the same format as the existing model explanations, based on the content from the following links.
https://aws.amazon.com/bedrock/anthropic/
https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html
| /** | ||
| * Mapping of AWS regions to their corresponding geographic areas for cross-region inference. | ||
| * This mapping is used to determine which cross-region inference profile to use based on the current region. | ||
| * This mapping is used to determine which cross-region inference profile to use based on the current region in prompt router. |
There was a problem hiding this comment.
The mapping here is used only in the prompt router.
Also, since the current implementation is sufficient for the base models supported by the prompt router, I only changed the docs.
Due to the expansion of cross-region inference grouping, the relationship between regions and groupings has become 1:n. For example, ap-northeast-1 is included in both APAC and JP.
Therefore, if the supported models for the prompt router increases in the future, adjustments to this implementation will likely be necessary.
| }, | ||
| ); | ||
| /**************************************************************************** | ||
| * ANTHROPIC |
There was a problem hiding this comment.
While there are other insufficient models besides Anthropic, there are too many to cover.
Therefore, I prioritized Claude, which is likely to be the most widely used.
mazyu36
changed the title
| * These regions help distribute traffic across multiple AWS regions for better | ||
| * throughput and resilience during peak demands. | ||
| */ | ||
| export enum CrossRegionInferenceProfileRegion { |
There was a problem hiding this comment.
Adding identifiers newly supported in Claude 4
https://docs.aws.amazon.com/bedrock/latest/userguide/inference-profiles-support.html
mazyu36
changed the title
aws-cdk-automation
dismissed
their stale review
✅ Updated pull request passes all PRLinter validations. Dismissing previous PRLinter review.
mazyu36
changed the title
aws-cdk-automation
added
the
pr/needs-community-review
2 participants
Issue # (if applicable)
N/A
Reason for this change
Although L2 Construct of bedrock-agentcore has been released, in the bedrock-alpha inference profiles that can be utilized within it, the definitions for Anthropic's latest Claude 4 Family and its associated cross-region inference profiles are missing.
Description of changes
Describe any new or updated permissions being added
N/A
Description of how you validated changes
Tests are not added because only enums are added.
Checklist
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache-2.0 license