Skip to content

[ENHANCEMENT] Add Prompt Caching Toggle to AWS Bedrock Settings #10576

@wisestmumbler

Description

@wisestmumbler

Problem (one or two sentences)

Users need the ability to toggle prompt caching when using AWS Bedrock.

The toggle should default to ON/checked and should be visible on the UI.

If feasible - Inference Profiles should also have an option in the settings.

Please reference:
#8669
#8670

https://aws.amazon.com/bedrock/cost-optimization/#:~:text=Use%20prompt%20caching%20to%20reduce,repeated%20prompt%20prefixes%20between%20requests.
https://aws.amazon.com/blogs/machine-learning/track-allocate-and-manage-your-generative-ai-cost-and-usage-with-amazon-bedrock/

Note:
This feature may already be partially implemented (#8670 (comment)) but users are unable to see the benefits

Context (who is affected and when)

Any user using AWS Bedrock would benefit from significant cost and latency reductions.
And have the confidence to know that prompt caching is on while having an escape hatch in case they need to turn it off

Desired behavior (conceptual, not technical)

In the providers settings when AWS Bedrock is chosen the option for prompt caching should be with the other checkboxes.

Of course, you should verify that prompt caching is working behind the scenes

Constraints / preferences (optional)

Ensure it works with most common AWS Bedrock setups, custom ARN, and govcloud.

Request checklist

  • I've searched existing Issues and Discussions for duplicates
  • This describes a specific problem with clear context and impact

Roo Code Task Links (optional)

No response

Acceptance criteria (optional)

No response

Proposed approach (optional)

No response

Trade-offs / risks (optional)

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    EnhancementNew feature or requestIssue/PR - TriageNew issue. Needs quick review to confirm validity and assign labels.

    Type

    No type

    Projects

    Status

    Done

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions