Amazon Bedrock tokenomics data lake
Web page · https://aws.amazon.com/blogs/machine-learning/amazon-bedrock-tokenomics-data-lake
Running Amazon Bedrock across many accounts shows you the bill but not what happened inside each call. This post builds a centralized observability data lake that aggregates Bedrock invocation logs from every linked account into one queryable table, so you can see prompts, token counts, model choice, and per-invocation cost across your organization.
First seen by Oriel on .
AI summary
AWS presents a deployable, centralized observability data lake that makes Amazon Bedrock invocations across accounts queryable and supports per-invocation token and cost analysis.
- CloudWatch subscription filters centralize invocation logs, Lambda cleans and flattens records, and Amazon Data Firehose converts them to partitioned Parquet files in Amazon S3.
- An AWS Glue table and Amazon Athena expose prompts, responses, token counts, model choices, stop reasons, timestamps, accounts, and IAM identities through SQL.
- Two AWS CloudFormation templates deploy the central pipeline and linked-account logging infrastructure, while model invocation logging must be enabled separately in each source account.
- Joining processed invocation data with CUR 2.0 using the IAM ARN attributes costs to prompts, teams, accounts, and inefficient calls.
- Centralizing full prompts and completions requires careful governance; AWS KMS encrypts stored data, while IAM and bucket policies restrict access to potentially sensitive information.
#amazon bedrock #observability #tokenomics #data lake
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| AWS Machine Learning Blog | fcb44d913a26d1a7b901b41bae672561748f42ff | https://aws.amazon.com/blogs/machine-learning/amazon-bedrock-tokenomics-data-lake/ Read discussion |