Part 2: Amazon Bedrock cost attribution with Amazon Athena and CUDOS

Building on the introduction of granular cost attribution for Amazon Bedrock, this article details how to set up Cost and Usage Report (CUR) 2.0 with IAM principal data, query Bedrock spend by principal, project, and team using Amazon Athena, and utilize CUDOS dashboards for detailed visualization. The setup involves creating a CUR 2.0 export with caller identity data enabled, which populates the line_item_iam_principal column and associated IAM principal tags. Once configured, users can connect the CUR data to Athena for SQL-based analysis, enabling detailed cost breakdowns by IAM principal, tags, and unknown schemas.

The article provides sample SQL queries for analyzing Bedrock costs by IAM principal and usage type, costs broken down by IAM principal tags, and dynamic tag discovery using Athena’s UNNEST function. These queries help organizations identify cost drivers, allocate expenses accurately, and optimize AI model usage.

Additionally, the article introduces the CUDOS dashboard, an open-source framework that offers pre-built visuals for Amazon Bedrock cost insights. The dashboard supports grouping by IAM principal, tags, models, and regions, with interactive drill-downs and trend analysis, facilitating data-driven cost management.

To get started, organizations should deploy CUDOS using the provided infrastructure templates, connect their CUR data to Athena, and run sample queries to verify setup. The article concludes with guidance on cleaning up resources and emphasizes the importance of ongoing cost monitoring for AI workloads.