Amazon SageMaker AI — Current Guide 2026
What changed?
AWS renamed Amazon SageMaker to Amazon SageMaker AI on December 3, 2024. The rename does not change existing SageMaker feature names or API namespaces. AWS also uses Amazon SageMaker as the broader next-generation platform for data, analytics, and AI capabilities.
So when discussing model development, training, and deployment, SageMaker AI is the precise current name.
Where SageMaker AI fits
SageMaker AI provides managed capabilities for building, training, tuning, deploying, and operating machine-learning and foundation-model workloads.
A production workflow commonly looks like:
Data → preparation → training/fine-tuning → evaluation → model artifact → deployment → monitoring → iteration
Notebook guidance
SageMaker still supports notebook instances, but they should not be confused with the older SageMaker Studio Classic experience.
For new work, prefer the current SageMaker Studio experience and current supported notebook platforms. AWS states that Studio Classic is maintained for existing workloads but is no longer available for new onboarding.
For notebook instances, AWS currently requires JupyterLab 4 for new instances and recommends the AL2023 platform. Amazon Linux 2 notebook instances reached end of support on June 30, 2026 and cannot be newly created or restarted from July 1, 2026.
Production checklist
- Use IAM least privilege and private networking where appropriate.
- Track data, model, and code versions.
- Separate training, evaluation, and production environments.
- Monitor latency, errors, throughput, model quality, and cost.
- Define rollback and recovery procedures before production deployment.
- Verify current service limits and pricing before committing to an architecture.
Connected knowledge
ML lifecycle → production ML platform → IAM least privilege → production incident checklist