# SageMaker ## AWS SageMaker — commitment brief > **Data sourced**: March 2026. Verify current figures at the [AWS Pricing Calculator](https://calculator.aws/pricing/2/home). ### Coverage summary Amazon SageMaker is covered by **SageMaker Savings Plans** (also called Machine Learning Savings Plans), which provide savings of up to 64% versus on-demand pricing in exchange for a consistent hourly spend commitment over a 1-year or 3-year term. SageMaker Savings Plans are the most flexible commitment instrument for ML workloads — a single plan automatically applies across Studio notebooks, processing jobs, training jobs, real-time inference, and more, regardless of instance type or region. There are no per-instance Reserved Instances for SageMaker. **What is covered**: SageMaker Studio notebooks, SageMaker On-Demand Notebooks, Processing, Training, Real-Time Inference (including Elastic Inference), Batch Transform, Data Wrangler, Feature Store (online store compute), Ground Truth (labeling jobs), and Managed Spot Training (the on-demand portion — Spot discounts apply separately to the Spot pricing layer). **What is not covered**: SageMaker Serverless Inference (pay-per-request; not eligible), SageMaker Canvas (no Savings Plan coverage), SageMaker JumpStart (no Savings Plan coverage for foundation model inference via JumpStart), SageMaker Edge Manager, model monitoring charges, and S3 storage used for model artifacts and datasets. ### Savings Plan types > Amazon SageMaker is covered by a single **SageMaker Savings Plan** type (also referred to as Machine Learning Savings Plans). The plan is a $/hour spend commitment that automatically applies to all eligible SageMaker usage across any instance type and any supported region. Terms: **1-year** (up to \~20–40% savings depending on payment option) and **3-year** (up to \~64% savings with All Upfront). Payment options: **All Upfront** (maximum discount), **Partial Upfront**, and **No Upfront** (\~20–41% savings). There are no EC2-style Standard or Convertible RI equivalents for SageMaker. ### Instance / usage type coverage | Usage Type | SageMaker Savings Plan | 1-yr (No Upfront) | 3-yr (All Upfront) | Notes | | ------------------------------------------- | ---------------------- | ----------------- | ------------------ | ----------------------------------------------------------- | | Studio Notebooks (ml.t3, ml.m5, ml.g4dn…) | ✅ Yes | \~20% | up to \~64% | All instance types covered by single plan | | On-Demand Notebooks | ✅ Yes | \~20% | up to \~64% | | | Processing Jobs (any instance type) | ✅ Yes | \~20% | up to \~64% | All ml.\* instance families covered | | Training Jobs (any instance type) | ✅ Yes | \~20% | up to \~64% | Includes GPU instances (ml.p3, ml.p4, ml.g5…) | | Real-Time Inference Endpoints | ✅ Yes | \~20% | up to \~64% | | | Batch Transform Jobs | ✅ Yes | \~20% | up to \~64% | | | Data Wrangler | ✅ Yes | \~20% | up to \~64% | | | Feature Store (online store) | ✅ Yes | \~20% | up to \~64% | | | Ground Truth Labeling | ✅ Yes | \~20% | up to \~64% | | | Managed Spot Training (on-demand component) | ✅ Yes | \~20% | up to \~64% | Spot discount on top of Savings Plan on on-demand component | | Serverless Inference | ❌ No | — | — | Pay-per-request; not eligible for Savings Plans | | SageMaker Canvas | ❌ No | — | — | No Savings Plan coverage | ## Regional availability **Commitment scope**: Region-flexible — SageMaker Savings Plans (Machine Learning Savings Plans) apply automatically across all eligible SageMaker usage types in any supported region; a single plan covers usage regardless of which region the workload runs in. Available in all major commercial AWS regions. SageMaker Savings Plans are supported across the primary US, Europe, and Asia Pacific regions, including GovCloud (US-West). Not all AWS regions support every SageMaker instance type — newer ML instance families (e.g., ml.p5, ml.trn1) may be available only in select regions. **Partition notes**: GovCloud (us-gov-west-1) supports SageMaker Savings Plans. Savings Plans are not available in China (cn-\*) regions. Not all SageMaker usage types or instance families are available in every region — verify the target region's SageMaker pricing page before committing. ⚠️ Regional and partition availability varies by instance type. Always verify at the [AWS Pricing Console](https://aws.amazon.com/ec2/pricing/) before purchasing. ### Archera Amazon SageMaker is a core service within Archera's commitment management scope. Archera can automate SageMaker Savings Plan purchases, monitor utilization across all covered SageMaker usage types, right-size the hourly spend commitment as ML workloads evolve, and wrap commitments in a GRI/GSP — eliminating downside risk on over-commitment while preserving the full Savings Plan discount. The cross-instance-type and cross-region flexibility of SageMaker Savings Plans makes them well-suited for Archera's automated commitment management. *** **Sources** * [Machine Learning Savings Plans](https://aws.amazon.com/savingsplans/ml-pricing/) * [SageMaker pricing](https://aws.amazon.com/sagemaker/pricing/) * [Decrease Your Machine Learning Costs with Instance Price Reductions and Savings Plans for Amazon SageMaker](https://aws.amazon.com/blogs/aws/slash-your-machine-learning-costs-with-instance-price-reductions-and-savings-plans-for-amazon-sagemaker/) * [Amazon SageMaker announces a price reduction and SageMaker Savings Plans with up to 64% cost saving](https://aws.amazon.com/about-aws/whats-new/2021/04/amazon-sagemaker-announces-a-price-reduction-in-instances-and-sagemaker-savings-plan/) * [What are Savings Plans?](https://docs.aws.amazon.com/savingsplans/latest/userguide/what-is-savings-plans.html) ⚠️ Discount percentages are approximate and usage-type/region-dependent. Always verify with the [AWS Pricing Calculator](https://calculator.aws/pricing/2/home).