Amazon SageMaker Updates

Amazon SageMaker AI inference now supports GPU partitioning for real-time endpoints

Amazon SageMaker AI inference now supports GPU partitioning, allowing multiple models to run on a single GPU with hardware-enforced isolation

Amazon SageMaker Unified Studio now supports custom designations

Amazon SageMaker Unified Studio introduces custom designations, allowing administrators to define permission sets aligned with organizational roles. This enables differentiated access for project members within a single project, eliminating the need for separate projects to isolate permissions.

SageMaker MLflow Apps now support MLflow 3.16

Amazon SageMaker AI MLflow Apps now support MLflow 3.16, offering a redesigned trace explorer and faster metric chart rendering for experiments with large metric histories

Amazon SageMaker Unified Studio adds light mode support for IDC-based domains

Amazon SageMaker Unified Studio now supports light mode for Identity Center (IDC) based domains, allowing users to choose between dark and light themes.

Amazon SageMaker AI now supports custom sequence lengths for serverless model customization jobs

Amazon SageMaker AI now allows users to configure sequence lengths in serverless model customization jobs, enabling control up to 16K tokens to better match training data characteristics.

Amazon SageMaker Studio simplifies getting started for new AI/ML builders with new sign-in options

Amazon SageMaker Studio now allows new customers to sign up with Google, GitHub, or Apple credentials and land in a fully configured environment-no AWS account setup or console navigation required.

Amazon SageMaker Unified Studio now supports two new connection capabilities: Iceberg REST Catalog connections and IAM authentication for Amazon DocumentDB

Amazon SageMaker Unified Studio adds support for Iceberg REST Catalog connections and IAM authentication for Amazon DocumentDB, enabling data teams to connect to more governed data sources and use credential-less authentication across projects

• Amazon DocumentDB

Amazon SageMaker HyperPod now supports model caching for faster inference autoscaling and reduced cold starts

Amazon SageMaker HyperPod now generally offers model caching, solving cold starts for LLM inference with up to 60% faster scale-out and 97% reduction in image-pull time

Amazon SageMaker Unified Studio now supports ODBC connections for Microsoft Power BI and other analytics tools

Amazon SageMaker Unified Studio adds ODBC support, enabling Microsoft Power BI and other ODBC-compatible tools to directly connect to governed data in SageMaker Unified Studio projects using the Amazon Athena ODBC driver

• Amazon Athena

Amazon SageMaker Feature Store now supports individual feature updates to lower write latency

Amazon SageMaker Feature Store adds support for updating individual features in a record, allowing data scientists to modify one or more feature values in a single request without rewriting the entire record, reducing write latency and cost

Amazon SageMaker AI Batch Transform Now Supports G6e Instances

Amazon SageMaker AI adds support for EC2 G6e instances in Batch Transform, enhancing performance for offline inference on large datasets

• Amazon EC2

Amazon SageMaker Unified Studio Workflows Supports Python and Bash Operators

Amazon SageMaker Unified Studio Workflows now supports PythonOperator and BashOperator, enabling custom Python functions and shell commands to run directly in serverless workflows

Amazon SageMaker Unified Studio CI/CD adds notebook promotion and AI-assisted manifest generation

Amazon SageMaker Unified Studio CI/CD now includes AI-assisted manifest generation and native notebook promotion across environments, accelerating project-to-production workflows

Generative AI Inference Recommendations now available in SageMaker AI Studio

Amazon SageMaker introduces automated inference recommendations in AI Studio, drastically reducing configuration time for generative AI models

Amazon SageMaker notebooks now support trusted identity propagation

Amazon SageMaker notebooks now support TIP, enabling per-user access control for data analytics

Amazon SageMaker Unified Studio now supports data profiling and anomaly detection

Amazon SageMaker Unified Studio adds data profiling and anomaly detection using AWS Glue Data Quality, enabling users to generate statistical profiles and detect data drift without predefined thresholds.

AWS Glue adds one-click access to SageMaker Unified Studio from the AWS console

AWS Glue now offers direct access to SageMaker Unified Studio, enabling data engineers and analysts to efficiently move from the Glue console to data querying, quality checks, and pipeline building with a single click

Amazon SageMaker Studio notebooks now support G7 instance types

Amazon SageMaker Studio notebooks now support G7 instances powered by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, delivering up to 4.6x AI inference performance over previous-generation G6 instances for tasks like conversational AI and video streaming.

Amazon SageMaker AI serverless model customization now supports full fine-tuning

Amazon SageMaker AI serverless model customization now supports full fine-tuning for over 25 open-source models, enabling deeper adaptation to domain-specific patterns and knowledge.

Amazon SageMaker Unified Studio now supports Teradata Vantage

Amazon SageMaker Unified Studio adds Teradata Vantage as a data source, enabling integration of enterprise data warehouse data with other assets for unified analysis

Amazon SageMaker Unified Studio brings richer Git version control to all project tools

Amazon SageMaker Unified Studio now provides full Git version control directly within tools like Query Editor, Visual ETL, Workflows, and Notebooks, enabling file-level management and flexible repository integration

SageMaker AI inference expands to G7e instances in new regions

Amazon SageMaker AI inference now offers G7e instances in Seoul, London, and Tokyo regions, supporting large LLM inference and high-memory workloads

Amazon SageMaker AI inference now supports G7 instances

Amazon SageMaker AI inference now supports G7 instances powered by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, delivering up to 4.6x AI inference performance over previous-generation G6 instances

SageMaker AI Inference Now Available with G6 Instances in AWS GovCloud (US-East)

Amazon SageMaker introduces G6 instances in AWS GovCloud (US-East), enabling government agencies and organizations to run generative AI workloads while meeting strict compliance and data residency requirements

Amazon SageMaker Unified Studio now supports Amazon OpenSearch

Amazon SageMaker Unified Studio now supports Amazon OpenSearch as a data source, allowing you to query and analyze search and log analytics data alongside other data assets.

Amazon SageMaker Unified Studio adds custom visual transforms

Amazon SageMaker Unified Studio now allows data engineers to create, share, and reuse custom visual transforms in visual ETL flows, enabling teams to apply business-specific logic without coding expertise.

Qwen3 embedding and reranking models for retrieval are now available in Amazon SageMaker JumpStart

Amazon SageMaker JumpStart now offers Qwen3-VL-Embedding-2B and Qwen3-Reranker-4B models for information retrieval and cross-modal understanding, enabling customers to build comprehensive search pipelines on AWS infrastructure.

Voxtral-Mini-4B-Realtime for real-time speech transcription now available in Amazon SageMaker JumpStart

Amazon SageMaker JumpStart now includes Mistral AI's Voxtral-Mini-4B-Realtime, a multilingual real-time speech transcription model supporting 13 languages with configurable latency and accuracy settings.

OpenAI privacy-filter for PII detection and masking now available in Amazon SageMaker JumpStart

Amazon SageMaker JumpStart now offers OpenAI's privacy-filter, a bidirectional token-classification model for detecting and masking personally identifiable information (PII) in text, enabling high-throughput data sanitization workflows.

gemma-4-E2B-it now available in Amazon SageMaker JumpStart

Amazon SageMaker JumpStart now offers Google DeepMind's gemma-4-E2B-it, a multimodal, instruction-tuned model optimized for efficient local execution. It processes text, image, and audio inputs, supports step-by-step reasoning, and provides capabilities like image understanding, video understanding, and code generation.

Amazon SageMaker HyperPod now supports AMI-based node lifecycle configuration for Slurm clusters using continuous provisioning

Amazon SageMaker HyperPod now supports AMI-based node lifecycle configuration for Slurm clusters using continuous provisioning, eliminating the need to download, configure, or upload lifecycle configuration scripts to Amazon S3.

Amazon SageMaker Feature Store now supports batch feature writes and record listing

Amazon SageMaker Feature Store adds batch feature writes and record listing capabilities, enabling scalable feature data processing and management

Amazon SageMaker Unified Studio adds custom asset types to the catalog in IAM-based domains

Amazon SageMaker Unified Studio now supports custom asset types for IAM-based domains, enabling cataloging of any asset format such as medical imaging files, PowerBI dashboards, or PDF reports

Amazon SageMaker now supports data lineage in IAM-based domains

Amazon SageMaker Unified Studio now supports OpenLineage compatible data lineage in IAM-based domains, capturing lineage events from Apache Spark jobs and providing interactive lineage graphs and a DeleteLineageEvent API

Amazon SageMaker Studio now integrates with Hugging Face for one-click model deployment and customization

Amazon SageMaker Studio now integrates directly with Hugging Face, allowing users to select a model and customize or deploy it within a pre-configured Studio environment in one click. New customers receive a ready-to-use environment with pre-configured permissions, while existing customers can access directly from Hugging Face.

Amazon SageMaker Unified Studio now supports Terraform for provisioning

Amazon SageMaker Unified Studio now supports Terraform for infrastructure-as-code deployment through version-controlled templates

Amazon SageMaker AI now supports serverless model customization for Gemma 4 models

Amazon SageMaker AI adds serverless model customization for Gemma 4 E4B and 31B models using supervised fine-tuning, direct preference optimization, and reinforcement fine-tuning

Amazon SageMaker AI cuts generative AI inference scale-out time by up to half with automatic container image caching

Amazon SageMaker Inference now supports container image caching, enabling up to 2x faster end-to-end scaling for generative AI models during scale-out events by pre-caching container images so new instances can start serving traffic faster without waiting to pull large images from Amazon ECR.

SageMaker Notebook Instances now support G6e instance types

Amazon SageMaker notebook instances now support EC2 G6e instances with NVIDIA L40s GPUs, enabling deployment of large language models and diffusion models for interactive training and fine-tuning.

Amazon SageMaker AI Announces New Observability Capability for Inference Endpoints

Amazon SageMaker AI provides new observability capabilities for inference endpoints, enabling real-time monitoring of token performance, GPU health, and infrastructure status to quickly diagnose and resolve issues.

Ministral-3-14B-Instruct for multimodal reasoning and agentic AI now available in Amazon SageMaker JumpStart

Amazon SageMaker JumpStart now offers the Ministral-3-14B-Instruct-2512 model, featuring multimodal capabilities and agentic functions in a compact 14B-parameter architecture optimized for edge deployment, supporting multiple languages.

all-MiniLM-L12-v2 for semantic search and sentence similarity now available in Amazon SageMaker JumpStart

Amazon SageMaker JumpStart now offers the all-MiniLM-L12-v2 model for semantic search and sentence similarity. This compact model provides fast inference with high-quality embeddings for information retrieval and text clustering applications

AWS Glue Interactive Sessions now supports Spark Connect for interactive workloads

AWS Glue Interactive Sessions now supports Apache Spark Connect, enabling users to develop and run Spark applications from their preferred environments on AWS Glue's serverless infrastructure without managing clusters

• AWS Glue

SageMaker AI now supports serverless fine-tuning for NVIDIA Nemotron models

Amazon SageMaker AI supports serverless fine-tuning for the NVIDIA Nemotron 3 Nano model, allowing users to customize the model with their proprietary data without managing infrastructure.

Amazon SageMaker Unified Studio Notebooks now support EMR Serverless

Amazon SageMaker Unified Studio Notebooks now support Amazon EMR Serverless, allowing data engineers and analysts to choose their Spark runtime for interactive analytics and data engineering workloads.