Pricing snapshot
Amazon SageMaker is priced through the AWS services used in the environment. SageMaker Unified Studio usage can include SageMaker Catalog charges, while storage, compute, and third-party services used through it carry their own pricing. SageMaker AI provides fully managed infrastructure, tools, and workflows to build, train, and deploy ML models.
Scroll sideways to see all columns. With a keyboard, focus this table and use the arrow keys.
| Pricing dimension | Published price or allowance | Billing note |
|---|---|---|
SageMaker Data Agent credit | $0.04 per credit | Credits are consumed by prompts and metered to the second decimal point; the minimum request consumption is 0.01 credits. |
SageMaker Catalog requests | $10 per 100,000 requests; 4,000 free requests per account in a billing month | Certain core APIs are free and do not count toward the request allowance. |
SageMaker Catalog metadata storage | $0.40 per GB; 20 MB free per account in a billing month | Charges apply when monthly free usage is exceeded. |
SageMaker Catalog compute | $1.776 per compute unit; 0.2 free compute units per account in a billing month | Compute is used for asset and schema metadata ingestion, and for automatic Git-to-S3 synchronizations only when you opt in for scheduled job runs. Billing applies to successfully completed ingestion, compute jobs, and synchronizations. |
Where to start
Use the Unified Studio Free Tier to explore always-free core requests and any remaining AWS Free Tier allocations or pay-per-use agreements attached to the AWS services you use.
Choose SageMaker notebooks when a serverless programming environment for analytics and machine learning jobs fits the work; notebook charges are based on the selected instance type and storage for the duration of use.
Use Data Agent when prompt-driven data querying, analysis, or model-development assistance is useful and per-credit billing is acceptable.
Free usage and billing conditions
Unified Studio's core requests—including manual IdC-based domain setup, project management, user management, and policy-grant management—are provided at no cost. AWS also states that Free Tier benefits are associated with the AWS account, so usage through other interfaces consumes the same allocation. For SageMaker notebooks, the stated Free Tier offer is 250 hours of sc.t3.medium notebook-instance use for the first two months.
The IdC quick-setup option can create additional networking-resource charges; AWS says the exact cost depends on account configuration. Manual setup is the option AWS identifies for full visibility during domain creation.
What drives the bill
SageMaker AI uses pay-as-you-go pricing without upfront commitments or minimum fees. Its listed cost dimensions include instance usage for training, hosting, and notebook instances; storage; data-processing jobs; model deployment; and MLOps components. Region, instance type, storage volume, and usage pattern can affect charges.
For SageMaker Catalog AI recommendations, AWS lists $0.015 per 1,000 input tokens and $0.075 per 1,000 output tokens, with no free usage for that dimension.
Published example
AWS gives a Data Agent example in which generating a complete data-transformation pipeline consumes 4 to 8 credits, described as 15 to 30 cents total depending on notebook context and catalog datasets. This is a vendor example, not a general workload estimate.
Take
For buyers using Unified Studio, budget at the service level: identify the AWS services a workflow will invoke, then track the relevant usage dimensions and free allowances. Data Agent and Catalog have published unit rates, while broader SageMaker AI costs depend on the resources and services selected.