AWS News - 2026-07-31

AWS Transit Gateway now supports Policy-Based Routing (PBR), giving network administrators granular control over how traffic is forwarded across their AWS network. With PBR, forwarding decisions can be based on a combination of packet attributes including source and destination IP addresses, ports, and protocol rather than destination IP address alone.

Previously, customers needing traffic steering or workload isolation had to build multi-VPC architectures with additional routing hops, adding complexity and operational overhead. PBR eliminates this by extending Transit Gateway's native routing capabilities, enabling security architects and enterprise network teams to classify and direct traffic inline without extra infrastructure. Customers associate a policy table with a Transit Gateway attachment and define an ordered set of rules. Each rule classifies traffic and directs matching packets to a specified route table using first-match-wins logic. This supports use cases such as steering sensitive workloads through AWS Network Firewall or third-party inspection appliances, routing application traffic over AWS Direct Connect or AWS VPN paths based on source, port, or protocol, and isolating production and development environments into separate routing domains to limit lateral movement.

Policy-Based Routing for AWS Transit Gateway is available in all commercial AWS Regions where Transit Gateway is available. You can configure PBR using the AWS Management Console, AWS Command Line Interface (CLI), and the AWS Software Development Kit (SDK). PBR incurs no additional charge beyond standard Transit Gateway fees. To learn more about Policy-Based Routing for AWS Transit Gateway, visit the AWS Transit Gateway product page .

On 7/30, OpenAI announced updated pricing for GPT-5.6 Terra and GPT-5.6 Luna. 

GPT-5.6 Terra is the balanced model for everyday production work, delivering GPT-5.5-level performance at lower cost. GPT-5.6 Luna is the fast, affordable model for high-volume inference tasks where latency and cost per token matter most. GPT-5.6 Sol pricing remains unchanged. Pricing on Amazon Bedrock matches OpenAI first-party rates, and usage counts toward your existing AWS commitments. 

GPT-5.6 Sol is available in US East (N. Virginia) and US East (Ohio). GPT-5.6 Terra and Luna are available in US East (N. Virginia), US East (Ohio), and US West (Oregon). For more details on the update, see the OpenAI blog. For the latest pricing information for GPT-5.6 models on Amazon Bedrock, please visit the Amazon Bedrock pricing page.

Today, OpenAI announced lower prices for GPT‑5.6 Luna and GPT‑5.6 Terra. Effective July 30, 2026, on-demand inference prices on Amazon Bedrock for GPT‑5.6 Luna are reduced by 80%, while prices for GPT‑5.6 Terra are reduced by 20%. These reductions are in-line with OpenAI's first-party pricing changes for these models. 

GPT‑5.6 Luna is optimized for fast, high-volume workloads and can use tools to complete multi-step workflows, making it well suited for content processing, classification, customer-service automation, and routine implementation tasks. GPT‑5.6 Terra balances intelligence, speed, and cost for everyday production workloads requiring more sophisticated reasoning. These price reductions enable customers to apply their capabilities across more applications, process larger workloads, and lower the cost per completed task. Pricing for GPT‑5.6 Sol remains unchanged. The new prices apply automatically, with no changes required from customers.

GPT‑5.6 Luna and Terra are available in US East (N. Virginia), US East (Ohio), and US West (Oregon) through the OpenAI Responses API on the bedrock-mantle endpoint. To get started, see the Amazon Bedrock OpenAI model documentation. For the latest pricing information for GPT-5.6 models on Amazon Bedrock, please visit the Amazon Bedrock pricing page.

Amazon Redshift now supports Graviton-based RG instances on the trailing track. Starting today, rg.large and rg.12xlarge instance types are available for customers running workloads on the trailing track on patch P202 and onwards.

The trailing track is designed for customers who prioritize stability for production workloads, running on a version already validated through the leading track. With RG instances now available on both tracks, customers can take advantage of AWS Graviton-powered performance - delivering up to 2.4x faster query performance than RA3 instances at 30% lower price per vCPU.

Customers on the trailing maintenance track (patch P202 and later) can now create Amazon Redshift RG clusters in all AWS regions where RG is generally available. To get started, customers can provision a new cluster or resize an existing cluster to an rg.large or rg.12xlarge instance type using the AWS Management Console, AWS CLI, or AWS SDKs.For more information, see Amazon Redshift cluster versions.

Amazon SageMaker Unified Studio gives project members full Git version control directly within the tools you already use - Query Editor, Visual ETL, Workflows, and Notebooks. The enhanced Repositories experience replaces the previous automatic sync approach with flexible, file-level version control. This brings a consistent source control experience across all project tools, including Notebooks, which previously had no Git support.

You choose exactly which files to track in Git by adding them to a repository on GitHub, GitLab, or Bitbucket - source control is not enforced at the project level, so you decide what gets versioned and when. When you're ready, you commit and push all your changes in a single action. Repositories are decoupled from project creation, meaning you can add a repository to a project at any point after the project is created, as your needs evolve. Projects can connect to any number of repositories and branches at the same time, and you can create branches, pull updates, and resolve conflicts without ever leaving your project. If you use JupyterLab or Code Editor, you also retain full Git CLI access through the built-in terminal. 

This feature is available in all AWS Regions where Amazon SageMaker Unified Studio is supported, for both IAM and IAM Identity Center domains. If your project uses the previous Git experience, you can opt in to the richer model by updating your project. To get started, see Working with repositories in the User Guide and Configuring Git connections in the Admin Guide.

Starting today, Amazon Elastic Compute Cloud (Amazon EC2) C7i instances powered by custom 4th Gen Intel Xeon Scalable processors (code-named Sapphire Rapids) are available in Europe (Milan) and Canada West (Calgary) regions. These custom processors, available only on AWS, offer up to 15% better performance over comparable x86-based Intel processors utilized by other cloud providers.

C7i instances deliver up to 15% better price-performance versus C6i instances and are a great choice for all compute-intensive workloads, such as batch processing, distributed analytics, ad-serving, and video encoding. C7i instances offer larger instance sizes, up to 48xlarge, and two bare metal sizes (metal-24xl, metal-48xl). These bare-metal sizes support built-in Intel accelerators: Data Streaming Accelerator, In-Memory Analytics Accelerator, and QuickAssist Technology that are used to facilitate efficient offload and acceleration of data operations and optimize performance for workloads. To learn more, visit the EC2 C7i instances Page.

Starting today, Amazon Elastic Compute Cloud (Amazon EC2) C7i-flex instances powered by custom 4th Gen Intel Xeon Scalable processors (code-named Sapphire Rapids) are available in Europe (Milan) region. These custom processors, available only on AWS, offer up to 15% better performance over comparable x86-based Intel processors utilized by other cloud providers.

C7i-flex instances are the easiest way for you to get price performance benefits for a majority of compute intensive workloads, and deliver up to 19% better price-performance compared to C6i. C7i-flex instances offer the most common sizes, from large to 16xlarge, and are a great first choice for applications that don't fully utilize all compute resources. With C7i-flex instances, you can seamlessly run web and application servers, databases, caches, Apache Kafka, and Elasticsearch, and more.

To learn more, visit the EC2 C7i-flex instances page.

AWS CodeDeploy is now available in five additional AWS Regions: Asia Pacific (New Zealand), Asia Pacific (Thailand), Asia Pacific (Taipei), Asia Pacific (Malaysia), and Mexico (Central). AWS CodeDeploy is a fully managed deployment service that automates application deployments to Amazon EC2 instances, on-premises servers, AWS Lambda functions, and Amazon ECS services. Customers in these newly added regions can now access CodeDeploy locally, enabling lower latency and supporting data residency requirements.

With AWS CodeDeploy, developers and DevOps engineers can rapidly release new features while avoiding downtime during application deployments. The service eliminates error-prone manual deployment operations and scales seamlessly from a single instance to thousands, making it ideal for teams managing complex deployment pipelines across EC2 fleets, serverless Lambda functions, or containerized ECS workloads.

AWS CodeDeploy is now available across 34 AWS commercial regions, as well as AWS GovCloud (US) and the AWS China Regions.

To learn more, visit the AWS CodeDeploy product page.

Amazon CloudWatch now supports collecting Prometheus metrics from your AWS infrastructure using fully managed collectors — enabling you to monitor Amazon EKS, Amazon EC2, Amazon ECS, Amazon MSK, and Amazon OpenSearch Service workloads without deploying or managing any agents.

Previously, getting Prometheus metrics into CloudWatch required deploying, scaling, and maintaining a self-managed OpenTelemetry Collector. Managed Prometheus collectors eliminate that overhead. You provide a scrape configuration and a connection to your resources, and CloudWatch handles provisioning, scaling, and collection automatically. Metrics are delivered in OpenTelemetry format and can be queried alongside your AWS vended metrics using PromQL — providing unified alarming, dashboarding, and cross-service correlation in a single view.

Managed collectors support Kubernetes service discovery (EKS), DNS-based service discovery via AWS Cloud Map (ECS), direct instance scraping (EC2), and open monitoring endpoints (MSK, OpenSearch). Metrics for EKS, MSK and Open Search can be visualized in automatic dashboards, queried with PromQL and used in CloudWatch alarms.

This feature is available in all AWS Regions where the CloudWatch OTLP endpoint is available, except Asia Pacific (New Zealand). Managed Prometheus collectors are charged by the hour and standard CloudWatch OpenTelemetry metric ingestion pricing applies. To get started, see the documentation.

AWS Lambda now supports Java 8, Java 11, and Java 17 runtimes on Amazon Linux 2023 (AL2023), available as both managed runtimes and container base images. These new runtimes enable customers running Lambda functions on Java 8, 11, or 17 to migrate from Amazon Linux 2 (AL2) to AL2023 without requiring a simultaneous Java version upgrade.

The existing Lambda runtimes for Java 8, 11 and 17 run on AL2, which reached end of life on June 30, 2026. To maintain support coverage and SLA eligibility, customers must migrate their functions to an AL2023-based runtime. AWS recommends that customers upgrade to Java 21 or Java 25 on AL2023 as the preferred migration path. These runtimes offer the latest language features and performance improvements. For customers who are not yet ready to upgrade their Java version, the new Java 8, 11, and 17 runtimes on AL2023 provide an alternative forward path. In addition, the existing AL2-based Lambda runtimes for Java 8, 11, and 17 will remain supported and continue to receive patches for critical and selected important security issues until June 30, 2027, giving customers a full year after AL2's end of life to complete their migration.

The Java 8, 11, and 17 runtimes on AL2023 are available in all AWS Regions where Lambda is available, including the AWS GovCloud (US) Regions and the China Regions.

To learn more about AL2023 support in Lambda, see our blog post. For customers ready to upgrade to Java 21 or Java 25, you can use AWS Transform custom to assist with the migration. To learn more about Lambda runtimes, visit the Lambda runtimes documentation.

Amazon Location Service now supports proximity-based point of interest discovery with GrabMaps data in the Asia Pacific (Singapore) and Asia Pacific (Malaysia) AWS Regions. Builders creating local discovery, delivery, and mobility applications in Southeast Asia can surface relevant nearby places using hyperlocal data validated by millions of daily Grab journeys.

SearchNearby returns places within a specified radius of a user's location, ranked by distance. Developers can filter results by category (such as fuel stations, hospitals, or restaurants) to narrow responses to what their application needs. This enables a delivery platform to find the closest pickup points, a ride-hailing app to show nearby dining options, or a travel app to locate services within range.

GrabMaps delivers hyperlocal mapping data purpose-built for Southeast Asia. Millions of daily journeys across the Grab platform continuously validate road networks, addresses, and points of interest, capturing changes in real time. GrabMaps covers detailed back alleys, narrow side streets, and motorcycle-accessible routes across more than 500 cities in Singapore, Malaysia, Thailand, Vietnam, Indonesia, Philippines, Myanmar, and Cambodia.

To get started, see the Search Nearby Developer Guide, API reference, and the GrabMaps Overview page.

NFS キャッシュソリューション KNFSD File Cache の Preview 提供開始、Amazon CloudWatch が AI コーディングエージェントの利用状況を可視化する Coding Agent Insights を発表、AWS CloudTrail でネットワークアクティビティイベントを ID 別に選択的にログ記録可能に、Amazon RDS for SQL Server が Microsoft SQL Server 2025 に対応、Amazon ECS がデプロイ中の操作を可視化する Action Logs を提供開始、Amazon SES が 3 段階の料金プランを導入、Amazon EKS の EKS Auto Mode と Karpenter で EFA と Placement Group をサポート、Network Load Balancer が送信元 IP アドレスタイプに基づくリスナールールに対応、Amazon CloudWatch Logs が Application Load Balancer ログに対応、AI エージェント向けオープンソースベンチマーク aws-bench をリサーチプレビューで公開、AWS で Claude Opus 5 が利用可能に等

NFS キャッシュソリューション KNFSD File Cache の Preview 提供開始、Amazon CloudWatch が AI コーディングエージェントの利用状況を可視化する Coding Agent Insights を発表、AWS CloudTrail でネットワークアクティビティイベントを ID 別に選択的にログ記録可能に、Amazon RDS for SQL Server が Microsoft SQL Server 2025 に対応、Amazon ECS がデプロイ中の操作を可視化する Action Logs を提供開始、Amazon SES が 3 段階の料金プランを導入、Amazon EKS の EKS Auto Mode と Karpenter で EFA と Placement Group をサポート、Network Load Balancer が送信元 IP アドレスタイプに基づくリスナールールに対応、Amazon CloudWatch Logs が Application Load Balancer ログに対応、AI エージェント向けオープンソースベンチマーク aws-bench をリサーチプレビューで公開、AWS で Claude Opus 5 が利用可能に等

メインフレーム基幹系システムの AWS 移行を真剣にご検討中のお客様向けの対面イベントを 2026/09/17(木) に麻布台ヒルズで開催します。参加費は無料です。詳細・お申し込みはブログをご覧ください。

Amazon Inspector is an automated vulnerability management service that continually scans Amazon Web Services (AWS) workloads for software vulnerabilities. The vulnerability management capabilities of Amazon Inspector are powered by an asset inventory engine known as the Amazon Inspector SBOM Generator (inspector-sbomgen), a standalone command-line tool that produces a software bill of materials (SBOM) from container […]

AI coding agents are part of the developer toolchain. Tools like Kiro and Claude Code generate features, tests, and code refactors from natural-language prompts. A single agent can open dozens of pull requests (PRs) across your repositories in an afternoon. That productivity comes with a trade-off: agents optimize for task completion at machine speed with […]

Amazon Bedrock Advanced Prompt Optimization optimizes your prompts for up to 5 models at once and compares original versus optimized performance across quality, latency, and cost. Migrate to a new model or improve your current one in minutes instead of weeks.

OpenAI GPT-5.6 Sol, Terra, and Luna are now generally available on Amazon Bedrock, along with explicit prompt caching that gives you precise control over which parts of your prompt are cached and reused. Learn how to get started, set up explicit caching, and migrate existing GPT workloads to reduce inference cost.

Learn how to build an inference meta-monitoring system for Amazon SageMaker AI endpoints using Amazon Quick. This governance layer sits above production ML inference pipelines to continuously track prediction and data quality, detect drift, integrate delayed ground truth, and surface automated performance dashboards.

In this post, we demonstrate how Yahoo implemented Amazon Bedrock to enhance their Search Retargeting (SRT) capabilities in the Yahoo DSP ad tech suite. SRT is a core audience targeting solution that helps advertisers reach users based on their historical search behavior, bridging search intent with display, video, and native advertising. Beyond targeting keywords entered on Yahoo Search, SRT uses AI to identify and engage users who demonstrate intent through search activity both on Yahoo and across integrated partner systems.

This post walks through deploying Kimi K3 on AWS using two approaches: Amazon SageMaker HyperPod, and  Amazon Elastic Kubernetes Service (Amazon EKS) cluster.

This post walks through deploying Kimi K3 on AWS using two approaches: Amazon SageMaker HyperPod, and  Amazon Elastic Kubernetes Service (Amazon EKS) cluster.