AWS adds usage-based billing for SageMaker bidirectional streaming
In short: Amazon has enabled usage-based software metering for bidirectional streaming models on SageMaker AI. Previously, software charges for long-running streaming sessions were difficult to capture, as they could not be reconciled with simple HTTP response headers. Model providers can now report usage incrementally via a dedicated metadata WebSocket connection.
This summary was generated automatically by AI from AWS Machine Learning's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1Introduced a companion metadata WebSocket for real-time reporting of billable units.
- 2Supports incremental billing reports during active sessions.
- 3Added header support for opt-in metering and optional strict connection requirements.
- 4Allows for negotiated per-unit pricing for streaming services via AWS Marketplace.
Why it matters
This update moves software billing away from hourly uptime models toward consumption-based pricing, which is critical for real-time voice and transcription services where session lengths vary significantly. It simplifies financial management by consolidating software and infrastructure charges directly into your AWS bill.
What it means for AI agents and contact centers
When evaluating or integrating third-party voice AI models on SageMaker, you can now expect transparent, per-unit billing. Test how your infrastructure handles the 'X-Amzn-SageMaker-Metadata-Stream-Required' setting to ensure that your inference availability is not compromised by potential metadata connection failures.
🧪 Worth Testing
Testing is advised for production environments relying on third-party models, as it allows for better cost control and integration with AWS Private Offers.
Sources
- AWS Machine LearningOfficialPrimary sourceOriginal article →„Introducing usage-based pricing for Amazon SageMaker AI bidirectional streaming“11 Nov 2026, 17:10
- Published by source
- 11 Nov 2026, 17:10
- Found by our system
- 11 Nov 2026, 17:33
- Summary generated
- 11 Nov 2026, 17:34
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy