Oriel — AI & technology worth noticing, every day.

Tokenomics at scale: Track and cap Amazon Bedrock spend per IAM principal with CUR 2.0 and AWS Budgets

Web page · https://aws.amazon.com/blogs/machine-learning/tokenomics-at-scale-track-and-cap-amazon-bedrock-spend-per-iam-principal-with-cur-2-0-and-aws-budgets

Get per-identity Amazon Bedrock cost visibility and enforceable spend limits using only AWS-native features, with no application code for the common case. Attribute spend with IAM principal tags, report with CUR 2.0 or Cost Explorer, and enforce with AWS Budgets actions that apply a Deny policy when a threshold is crossed.

First seen by Oriel on .

AI summary

This AWS guide presents a native approach to attribute Amazon Bedrock costs by IAM identity, report them through Cost Explorer or CUR 2.0, and enforce model-specific spending limits with AWS Budgets.

  • Amazon Bedrock records the IAM user or role for each inference request, while principal or session tags add user, team, and cost-center dimensions.
  • Cost allocation tags enable Cost Explorer reporting; CUR 2.0 with caller identity and Amazon Athena supports detailed SQL analysis of aggregated billing data.
  • CUR queries must include input, output, cache-read, and cache-write tokens and match pricing by token type, service tier, and cross-Region routing.
  • AWS Budgets can attach Deny IAM policies at thresholds, including tiered restrictions, but shared roles cause the action to affect every user using them.
  • Budget data refreshes up to three times daily, tags can take up to 48 hours to populate, and CUR does not provide per-request line items.

#amazon bedrock #aws budgets #cost allocation #iam

Generated by gpt-5.6-sol. Check the original for details.

Aliases

SourceExternal IDURL
AWS Machine Learning Bloge0c369a0dd940a1e529c2070289fe5f5a34b16a2https://aws.amazon.com/blogs/machine-learning/tokenomics-at-scale-track-and-cap-amazon-bedrock-spend-per-iam-principal-with-cur-2-0-and-aws-budgets/ Read discussion