跳到正文

如何按 IAM 主体跟踪并限制 Amazon Bedrock 支出

AWS Machine Learning Blog当事人· Arun Chandapillai·· 2 天前Tokenomics at scale: Track and cap Amazon Bedrock spend per IAM principal with CUR 2.0 and AWS Budgets
打开原文
原文
图片来自 AWS Machine Learning Blog
谁
AWS
引文
This post shows how to get per-identity Amazon Bedrock cost visibility and enforceable spend limits using only AWS-native features, with no application code to write for the common case.

发生了什么

AWS 作者介绍了如何按 IAM 用户或角色跟踪 Amazon Bedrock 成本,并使用成本分配标签、Cost Explorer 或 CUR 2.0 以及 AWS Budgets 强制执行支出限额。操作指南涵盖主体标签和会话标签、用于详细报告的 Athena 查询,以及会附加 Deny IAM 策略的预算操作。预算数据每天最多刷新三次,而共享角色的 Deny 操作会阻止所有使用该角色的人。

接下来读什么

查看这件事的完整进展AWS Details Cost Attribution and Spending Controls for AI Tools on Amazon Bedrock

这件事到目前为止

AWS explains how to attribute Amazon Bedrock inference costs to individual users of AI coding and productivity tools, then compare ways to control their usage. The guide configures IAM Identity Center session identities and tags for CUR 2.0 reporting, with Athena queries and dashboard options for analyzing per-user costs. It also compares AWS Budgets, CloudWatch alarms with Lambda, DynamoDB quotas, and LLM gateways as spending controls. AWS’s earlier walkthrough covers tracking Bedrock costs by IAM user or role with cost allocation tags, Cost Explorer or CUR 2.0, and AWS Budgets actions that attach a Deny policy. Budgets refresh up to three times a day, so spending can exceed a threshold before an action runs. Newly applied cost allocation tags can take up to 24 hours to appear and another 24 hours after activation to populate billing data. A Deny policy on a shared role blocks everyone using that role.