NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #882 most downloaded on PyPI
Open source library for training and deploying models on Amazon SageMaker.
Last release 10 days ago
24 Sep 2026
Ships on a steady schedule
a new release about every 2 weeks
Nearly every release is documented
notes for 59 of the last 60 stable releases
3 versions withdrawn
withdrawn after publishing
9 years old
692 releases · first in 2017
Add support for Streaming Inference
Update SM Python SDK for PT 2.2.0 SM DLC
One column per quarter.
pin dll version to support python3.11 to the sdk
tolerate vulnerable old model for integ test and temporarily skip test_list_jumpstart_models_script_filter
ModelBuilder to fetch local schema when no SchemaBuilder present.
Support collection type and target store for feature store ingestion.
Add PT 2.1 as a supported framework for the smdistributed distribution
Introduce HF Transformers to ModelBuilder
Logic to detect hardware GPU count and aggregate GPU memory size in MiB
Support selective pipeline execution for function step
add throughput management support for feature group
update image_uri_configs 12-29-2023 06:17:34 PST
support local mode in SageMaker Studio
get_execution_roleupdate image_uri_configs 12-22-2023 06:17:35 PST
support remote debug for sagemaker training job
Use specific images for SMP v2 jobs
Merge branch 'master-rba' into local_merge
remove two setuptools deprecations
partitions key requirement on pysdk side.Session.download_data can not download nested objectsGoldfinch InferenceComponent integration
update get_execution_role_arn from metadata file if present
### Features * re:Invent 2023 keynote1
update get_execution_role_arn from metadata file if present
PT2.1 SM Training/Inference DLC Release
inference instance type conditioned on training instance type
Allow either instance_type or instance_group to be defined in…
jumpstart gated model artifacts
Added register step in Jumpstart model
jumpstart model artifact instance type variants
v2.7.2import error in unsupported js regions
jumpstart estimator enable infra check flag
Selective Step Execution milestone 2 features
Add support for in-memory feature groups and collection type features in Feature Store.
add feature processor APIs to public doc
jumpstart instance type variants
SkipModelValidation in modelRegistry
### Features * TGI 1.0.3 Image URI Config
Local Mode - Add Support for Docker Compose V2
### Documentation Changes * add interactive apps rst file
Enable notebook instances to get presigned url
Update pytorch.json with 2.0.1 for inference and training
image url for modelmonitor in TLV region
temporarily skip kmeans notebook
Add presigned URLs for interactive apps
attach method for jumpstart estimator
Support to get latest monitoring execution processing logs
chore: excessive jumpstart bucket logging
Add TLV accounts for 1P Algorithms
Add TF 2.13 Training and Inference SM images
Add huggingface-llm 0.9.3 dlc images
chore: jumpstart deprecation messages
Update the apache airflow constraints
Add check for if TrialComponent is already associated with a Trial in Run
Add PipelineDefinitionConfig to pipelines to toggle custom job …
protobuf3-to-dictEnable customizing artifact output path
Add support for tags in to_pipeline API for feature processor
Support uncompressed model upload
add SageMaker FeatureStore feature processing
Remove deprecated option.s3url in favor of option.model_id
inf2 support to HuggingFaceModelAdd support for Deployment Recommendation ID in model.deploy(). No tagging support
SDK Defaults - Environment Variables
Your coding agent can read these notes before it upgrades. Set up the MCP server →