AWS releases
Amazon Web Services releases and Terraform AWS provider. New features, breaking changes, security advisories and deprecations - each summarised in plain English and updated continuously.
Tracking 284 AWS releases · Updated
- AWS What's New aiawspreviewengineer ·
AWS Blocks: Open-source framework for application backends (Preview)
AWS Blocks, an open-source TypeScript framework, is now in public preview, simplifying backend development on AWS by abstracting infrastructure tools. It allows developers to run a local environment with essential services and deploy the same code to production AWS services without modification. This framework aims to accelerate the development of applications by integrating database, authentication, and AI capabilities with full-stack type safety and built-in guidance for AI coding tools. It supports various frontend frameworks and is available at no extra charge, with deployment across all commercial AWS regions.
feature announcement - AWS What's New aiawsengineer ·
AWS Marketplace Adds AI Assistant for Product Listings
AWS Marketplace now offers an AI-assisted product listing feature within Partner Assistant chat. This capability helps ISVs and Consulting Partners automatically generate and optimize listings using existing digital assets, reducing manual effort and guesswork. It provides field-level recommendations and quality scores to improve discoverability by buyers. The feature is available in Partner Central and AMMP, excluding GovCloud and China Regions.
feature announcement - AWS What's New aiawsengineer ·
AWS Partner Central agents now offer real-time deal qualification and recommendations
AWS Partner Central agents have been enhanced to qualify co-sell opportunities in real time and provide recommendations that accelerate deal progression. This new functionality allows agents to enrich opportunity details conversationally, eliminating manual review delays for partners. The system now matches opportunities to specific co-sell motions and assigns an Opportunity Quality Score to guide AWS engagement, with the enhanced experience available today to AWS Partners globally.
feature - AWS What's New aiawsmedia ·
xAI's Grok 4.3 available on Amazon Bedrock
Amazon Bedrock now offers xAI's Grok 4.3 model, expanding choices for generative AI applications. Grok 4.3 is a reasoning-first model designed for consistent multi-step agent behavior and efficient token usage, making it suitable for enterprise workloads like contract review and financial analysis. It runs on Bedrock's new Mantle inference engine, optimized for performance and cost.
feature announcement - AWS What's New aiawsengineerfinancemedia ·
AWS WAF adds AI traffic monetization for bots
AWS WAF now offers AI traffic monetization, a new capability allowing content owners to set prices, accept payments, and grant access to AI bots accessing their APIs and content. This feature utilizes an HTTP 402 response for machine-to-machine payments, with initial integration via Coinbase and upcoming support for Stripe. It enables differentiated pricing and revenue analytics, and is available at no additional charge for AWS WAF customers.
feature - AWS What's New aiawsgaengineer ·
AWS Bedrock AgentCore Memory Adds Strictly Consistent Metadata
Amazon Bedrock AgentCore Memory now allows developers to directly attach metadata to long-term memory records, ensuring data integrity and enabling precise filtering. This feature improves data organization, facilitates compliance, and supports multi-tenant architectures by guaranteeing metadata values are preserved without LLM inference. Strictly consistent metadata is available in all AWS Regions where AgentCore Memory is supported.
feature - AWS What's New aiawsengineer ·
SageMaker now offers serverless fine-tuning for NVIDIA Nemotron models
Amazon SageMaker now supports serverless fine-tuning for NVIDIA's Nemotron 3 Nano model, allowing users to adapt foundation models with proprietary data. This feature simplifies the process by managing infrastructure and orchestration, enabling users to focus on data and evaluation. Serverless customization is now available in specific AWS regions and can be accessed via SageMaker Studio or the Python SDK.
feature - AWS What's New aiawsengineer ·
SageMaker Serverless Fine-Tuning for Nvidia Nemotron Models
Amazon SageMaker now offers serverless fine-tuning for Nvidia's Nemotron 3 Nano model, allowing users to adapt foundation models with proprietary data. This feature enables tailoring models for domain-specific tasks or tone alignment, abstracting infrastructure management and adopting a pay-for-use model. It's available in specific AWS regions and can be accessed via SageMaker Studio or the Python SDK.
feature - AWS What's New aiawsgaengineergovernment ·
Amazon EC2 Capacity Blocks for ML available in AWS GovCloud
Amazon EC2 Capacity Blocks for ML are now available in AWS GovCloud (US) Regions, offering reserved GPU capacity for government and regulated industries. This feature allows customers to secure GPU instances for ML workloads like training and prototyping up to eight weeks in advance for durations up to six months. The service provides low-latency connectivity and can be shared across accounts via AWS Resource Access Manager (RAM). This enablement is crucial for organizations needing assured, on-demand accelerated compute within the stringent compliance requirements of AWS GovCloud.
feature announcement - AWS What's New aiawsgaengineer ·
Amazon Bedrock AgentCore adds Web Search
Amazon Bedrock AgentCore now offers general availability for Web Search, a fully managed tool to ground AI agents in current web knowledge. This eliminates the need for custom integrations with external search providers, simplifying agent development and enhancing response accuracy with multi-source grounding. The feature is available today in the US East (N. Virginia) region and is optimized for token-efficient retrieval of relevant information.
feature announcement - AWS What's New aiawssnowflakeengineer ·
Amazon Quick integrates with Snowflake Cortex AI
Amazon Quick now integrates with Snowflake Cortex AI via the Model Context Protocol, allowing users to query Snowflake data and documents using natural language and automate multi-step workflows directly within Quick. This integration enables teams to build orchestrated workflows with Snowflake Cortex Agents for repeatable, governed processes. It benefits any multi-step process involving structured data and unstructured documents, offering both structured rigor and conversational flexibility.
feature announcement - AWS What's New aiawsgamedia ·
Google Gemma 4 models available on Amazon Bedrock
AWS has announced the integration of Google DeepMind's Gemma 4 family of open-weight models into Amazon Bedrock. These models enable the development of generative AI applications for reasoning, multimodal understanding, agentic workflows, and software engineering. The Gemma 4 family includes three variants optimized for different use cases, all running on a new, price-performant Bedrock innovation. The models are now available in select AWS regions.
feature announcement - AWS What's New aiawsmedia ·
OpenAI GPT-5.4 and GPT-5.5 models now on Amazon Bedrock
AWS has expanded the availability of OpenAI's GPT-5.4 and GPT-5.5 models to the US East (N. Virginia) Region on Amazon Bedrock. These models offer advanced reasoning, coding, and agentic task capabilities for building generative AI applications. GPT-5.5 is OpenAI's most capable model for complex tasks, while GPT-5.4 supports frontier reasoning and long-context workflows. Both support a 272K-token context window with text and image inputs and are available via the Responses API.
feature announcement - AWS What's New aiawsgaengineer ·
Amazon EC2 P6-B200 instances with NVIDIA Blackwell GPUs now in AWS GovCloud (US-East)
Amazon EC2 P6-B200 instances, powered by NVIDIA Blackwell GPUs, are now available in the AWS GovCloud (US-East) Region. These instances deliver up to twice the performance of P5en instances for AI training and inference workloads, featuring advanced GPUs, increased memory bandwidth, and enhanced networking capabilities. They are designed for secure and scalable AI workloads within Amazon EC2 UltraClusters and are also available in other US regions.
feature announcement - AWS What's New aiawsgapreviewengineer ·
AWS announces Claude Fable 5, a generally available AI model
AWS has made Claude Fable 5, a state-of-the-art Mythos-class AI model, generally available. This model offers enhanced autonomous knowledge work and coding capabilities for developers and enterprises building AI applications. Claude Fable 5 is designed for professional tasks across various industries and can be accessed via Amazon Bedrock or the Claude Platform on AWS. Amazon Bedrock offers AWS-managed features and data residency, while the Claude Platform provides direct Anthropic access with unified billing.
feature announcement - AWS What's New aiawsengineer ·
AWS Cost Explorer adds Amazon Q-powered cost analysis
AWS Cost Explorer now integrates with Amazon Q Developer to provide automated cost explanations for any report. This feature analyzes cost trends, drivers, and anomalies within the context of user-defined filters and time periods, reducing manual investigation effort. It is available in all commercial AWS Regions at no additional charge and allows for follow-up questions to explore insights.
feature - AWS What's New aiawsengineer ·
Amazon Connect adds AI agent trace details for voice interactions
Amazon Connect now offers AI agent traces, providing detailed insights into how AI agents reason and act during customer voice interactions. This enhancement allows users to understand AI decision-making, diagnose issues, and build agentic experiences with greater confidence. The feature is available in all AWS Regions supporting Amazon Connect Customer AI Agents and is accessible via the Connect web UI alongside conversation transcripts.
feature - AWS What's New aiobservabilityawsengineer ·
AWS Cost Anomaly Detection adds AI-powered root cause investigation
AWS Cost Anomaly Detection now integrates Amazon Q to automatically analyze the root cause of cost anomalies, delivering plain-language explanations in minutes instead of hours. This feature aids FinOps and engineering teams by identifying usage or rate drivers, contributing services, accounts, regions, and specific API calls. Available immediately in all commercial AWS Regions at no extra charge, it streamlines the transition from alert to action for cost optimization efforts.
feature - AWS What's New aiawsengineer ·
Amazon Bedrock AgentCore Runtime adds interactive terminal shells
Amazon Bedrock AgentCore Runtime now supports interactive shells via a new API, providing a persistent, PTY-backed terminal directly into agent sessions over WebSocket. This feature is crucial for developers hosting coding agents, enabling them to interact, inspect files, and debug within an isolated microVM environment. It offers a full terminal experience, including state persistence, automatic reconnection, and support for up to 10 concurrent shells per runtime.
feature - AWS What's New aiawsengineer ·
AWS CLI adds support for Agent Toolkit installation wizard
The AWS Command Line Interface (CLI) now supports an interactive wizard for installing the Agent Toolkit for AWS across multiple coding agents, simplifying setup and management. This enhances agent capabilities by providing access to AWS APIs and expert guidance through skills and plugins. The Agent Toolkit's MCP Server is now available in US East (N. Virginia) and Europe (Frankfurt).
feature announcement