Releasesby Glyph
Amazon SageMaker AI inference now supports GPU partitioning for real-time endpoints
SageMaker AI now supports GPU partitioning using NVIDIA Multi-Instance GPU (MIG) technology. This allows multiple models to share a single GPU with hardware-enforced isolation.
Also shipped Oct 30, 2026AWS in October 2026 →
Microsoft Graph Logs connector is available in Beta
The Microsoft Graph Logs connector is now in beta for Databricks Lakeflow Connect, supporting incremental ingestion of security alerts from Microsoft security products.
Hybrid table support for Google Cloud (Preview)
Snowflake hybrid tables are now supported in all Google Cloud commercial regions in preview.
External OAuth for Snowflake Postgres (General availability)
External OAuth for Snowflake Postgres is now generally available, allowing authentication via external identity providers.
Spanner supports equality and ordering comparisons on JSONB and JSON values
Spanner now allows equality and ordering comparisons on JSONB and JSON values directly in GROUP_BY and ORDER BY clauses. It also supports implicit type coercion and search indexes for JSON equality operations.
[In preview] Public Preview: Support database-level managed identity (including FIC) in Create Database Scoped Credential T-SQL
Users can now assign a user-assigned managed identity to individual databases in Azure SQL Database. This enables secure access to files in Azure Blob Storage and Data Lake Storage Gen2 via external tables and OPENROWSET.
Inference request metadata logging (GA)
Inference request metadata logging is now generally available for Gemini and Claude models. It records model usage, latency, and error codes without capturing prompt or response content.
More from Amazon SageMakerEvery Amazon SageMaker release →
Amazon SageMaker AI now supports SOCI-indexed images to start training jobs in seconds
Amazon SageMaker AI now supports Seekable OCI (SOCI)-indexed container images, allowing training and processing jobs to lazy-load images and start in seconds instead of minutes.
Amazon SageMaker Unified Studio now supports custom designations
Amazon SageMaker Unified Studio now supports custom designations, named permission sets that control what actions individual project members can perform. Previously, project members could only be assigned Owner or Contributor designations, which provided nearly identical permissions.
SageMaker MLflow Apps now support MLflow 3.16
Amazon SageMaker MLflow Apps now support MLflow 3.16, featuring a redesigned trace explorer and improved experiment tracking performance for generative AI and agentic workflows.
Amazon SageMaker Unified Studio adds light mode support for IDC-based domains
Amazon SageMaker Unified Studio now supports a light mode theme for Identity Center-based domains to improve accessibility.
Amazon SageMaker AI announces training session APIs for serverless model customization
New training session APIs for SageMaker AI serverless model customization allow developers to define custom training loops for SFT, DPO, and reinforcement learning without managing GPU infrastructure.
Amazon SageMaker AI now supports serverless model customization for Muse Glimmer, Qwen 3.8 and Ministral 3 models
SageMaker AI now supports serverless model customization for Muse Glimmer, Qwen 3.8, and Ministral 3 models. Users can apply SFT, DPO, and RFT techniques using LoRA or full fine-tuning.
Amazon SageMaker AI now supports custom sequence lengths for serverless model customization jobs
Today, Amazon SageMaker AI announces the ability to configure the sequence length used during your serverless model customization job. Amazon SageMaker AI serverless model customization supports fine-tuning for many open-weight models, including models from popular families including gpt-oss, Gemma, Llama, Nemotron, and Qwen.
Nemotron 3.5 Super VL model now available on Amazon SageMaker JumpStart
NVIDIA's Nemotron 3.5 Super VL is now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. This vision-language model from the Nemotron 3.5 family accepts images, video, and text as input and generates text, enabling customers to deploy multimodal AI solutions on AWS infrastructure.