Features
Real-time experiment tracking
Monitor training and model behavior
Compare thousands of runs
Analyze metrics across layers
Surface issues during training
Model registry and lineage
Deep integration into training stacks
SSO and on-prem deployment (Enterprise)
Unlimited tracking (Team tier)
200 hours/month free tracking
SageMaker AI for full ML lifecycle (build, train, deploy)
SageMaker Unified Studio for analytics and AI development
Amazon SageMaker Catalog for data and AI governance
Lakehouse architecture unifying S3, Redshift, and federated sources
Zero-ETL integrations for near real-time data ingestion
HyperPod for distributed training of large models
JumpStart for pre-built models and solutions
MLOps tools for model management and versioning
Amazon Q Developer for natural language productivity
Federated query across third-party data sources
Fine-grained access controls and permissions
Built-in data quality monitoring and lineage
Apache Iceberg compatibility for open table formats
Bedrock integration for generative AI applications
Web Search on Bedrock AgentCore for grounded agent responses
