Today, AWS Deadline Cloud announces support for configurable job scheduling modes, giving you control over how workers are distributed across jobs in a queue. AWS Deadline Cloud is a fully managed service that simplifies render management for computer-generated 2D/3D graphics and visual effects for films, TV shows, commercials, games, and industrial design.
Previously, all available workers were assigned to the highest-priority, earliest-submitted job first, which could delay feedback on other submitted jobs. You can now choose from three scheduling modes when creating or updating a queue: priority FIFO (the existing default behavior), priority balanced (workers are distributed evenly across all jobs at the highest priority level), and weighted balanced (jobs are weighted based on configurable parameters including priority, error count, submission time, and rendering task count). Priority balanced and weighted balanced scheduling modes enable artists to get immediate feedback on their submissions without waiting for earlier jobs to complete.
Configurable job scheduling modes are available in all AWS Regions where AWS Deadline Cloud is supported. To get started, visit the Deadline Cloud developer guide.
Today, AWS Deadline Cloud announces support for configurable job scheduling modes, giving you control over how workers are distributed across jobs in a queue. AWS Deadline Cloud is a fully managed service that simplifies render management for computer-generated 2D/3D graphics and visual effects for films, TV shows, commercials, games, and industrial design. Previously, all available workers were assigned to the highest-priority, earliest-submitted job first, which could delay feedback on other submitted jobs. You can now choose from three scheduling modes when creating or updating a queue: priority FIFO (the existing default behavior), priority balanced (workers are distributed evenly across all jobs at the highest priority level), and weighted balanced (jobs are weighted based on configurable parameters including priority, error count, submission time, and rendering task count). Priority balanced and weighted balanced scheduling modes enable artists to get immediate feedback on their submissions without waiting for earlier jobs to complete. Configurable job scheduling modes are available in all AWS Regions where AWS Deadline Cloud is supported. To get started, visit the Deadline Cloud developer guide.
Amazon Lightsail now offers compute-optimized instance bundles with up to 72 vCPUs. The new instance bundles are available in 7 sizes with both IPv6-only and dual-stack networking types. All Lightsail blueprints are supported with compute-optimized instance bundles, including Linux and Windows operating system (OS) and application blueprints. You can create instances using the new bundles with pre-configured OS and application blueprints including WordPress, cPanel & WHM, Plesk, Drupal, Magento, MEAN, LAMP, Node.js, Ruby on Rails, Amazon Linux, Ubuntu, CentOS, Debian, AlmaLinux, and Windows.
The new compute-optimized instances enable you to run compute-intensive workloads that require high CPU. These high-performance instances deliver consistent, dedicated CPU performance ensuring your applications always have the full processing power they need. These new instance bundles are ideal for workloads such as batch processing, distributed analytics, high-performance web servers, scientific modeling, dedicated gaming servers, ad serving engines, video encoding, and CPU-intensive machine learning inference applications.
Amazon Lightsail is available in 15 AWS Regions including US East (N. Virginia), US West (Oregon), Europe (Frankfurt), Europe (London), Asia Pacific (Tokyo), and Asia Pacific (Jakarta). To get started, visit the Lightsail console. For pricing and other details, visit the Amazon Lightsail pricing.
Amazon Lightsail now offers compute-optimized instance bundles with up to 72 vCPUs. The new instance bundles are available in 7 sizes with both IPv6-only and dual-stack networking types. All Lightsail blueprints are supported with compute-optimized instance bundles, including Linux and Windows operating system (OS) and application blueprints. You can create instances using the new bundles with pre-configured OS and application blueprints including WordPress, cPanel & WHM, Plesk, Drupal, Magento, MEAN, LAMP, Node.js, Ruby on Rails, Amazon Linux, Ubuntu, CentOS, Debian, AlmaLinux, and Windows.
The new compute-optimized instances enable you to run compute-intensive workloads that require high CPU. These high-performance instances deliver consistent, dedicated CPU performance ensuring your applications always have the full processing power they need. These new instance bundles are ideal for workloads such as batch processing, distributed analytics, high-performance web servers, scientific modeling, dedicated gaming servers, ad serving engines, video encoding, and CPU-intensive machine learning inference applications.
Amazon Lightsail is available in 15 AWS Regions including US East (N. Virginia), US West (Oregon), Europe (Frankfurt), Europe (London), Asia Pacific (Tokyo), and Asia Pacific (Jakarta). To get started, visit the Lightsail console. For pricing and other details, visit the Amazon Lightsail pricing.
Amazon WorkSpaces Applications now enables instances in multi-session fleets to stop accepting new user sessions, while allowing existing sessions to continue uninterrupted. This capability, known as drain mode, ensures seamless operations during maintenance, scaling, or system updates.
Multi-session fleets allow hosting multiple end user sessions on a single instance, helping to maximize usage of the underlying infrastructure – including compute, memory and storage resources – and lowering the overall cost. This new drain mode capability helps administrators manage multi-session environments more seamlessly by preventing disruption to active users. When performing system maintenance, applying security patches, or scaling down resources, administrators can configure instances to gradually empty without abruptly terminating user sessions. This ensures users can complete their work smoothly while new connections are directed to other available instances, maintaining system stability and improving the overall end-user experience.
This new feature is available at no additional cost in all AWS Regions where Amazon WorkSpaces Applications is available.
Amazon WorkSpaces Applications now enables instances in multi-session fleets to stop accepting new user sessions, while allowing existing sessions to continue uninterrupted. This capability, known as drain mode, ensures seamless operations during maintenance, scaling, or system updates. Multi-session fleets allow hosting multiple end user sessions on a single instance, helping to maximize usage of the underlying infrastructure – including compute, memory and storage resources – and lowering the overall cost. This new drain mode capability helps administrators manage multi-session environments more seamlessly by preventing disruption to active users. When performing system maintenance, applying security patches, or scaling down resources, administrators can configure instances to gradually empty without abruptly terminating user sessions. This ensures users can complete their work smoothly while new connections are directed to other available instances, maintaining system stability and improving the overall end-user experience. This new feature is available at no additional cost in all AWS Regions where Amazon WorkSpaces Applications is available. Amazon WorkSpaces Applications offers pay-as-you-go pricing. For more information, see Amazon WorkSpaces Applications Pricing. To get started with WorkSpaces Applications, see WorkSpaces applications: Getting started.
Today, AWS announced the expansion of 100 Gbps dedicated connections at the existing AWS Direct Connect location in the Datacom Orbit DH6 data center near Auckland, New Zealand. You can now establish private, direct network access to all public AWS Regions (except those in China), AWS GovCloud Regions, and AWS Local Zones from this location. This is the second AWS Direct Connect location in New Zealand to provide 100 Gbps connections with MACsec encryption capabilities.
The Direct Connect service enables you to establish a private, physical network connection between AWS and your data center, office, or colocation environment. These private connections can provide a more consistent network experience than those made over the public internet.
For more information on the over 150 Direct Connect locations worldwide, visit the locations section of the Direct Connect product detail pages. Or, visit our getting started page to learn more about how to purchase and deploy Direct Connect.
Today, AWS announced the expansion of 100 Gbps dedicated connections at the existing AWS Direct Connect location in the Datacom Orbit DH6 data center near Auckland, New Zealand. You can now establish private, direct network access to all public AWS Regions (except those in China), AWS GovCloud Regions, and AWS Local Zones from this location. This is the second AWS Direct Connect location in New Zealand to provide 100 Gbps connections with MACsec encryption capabilities.
The Direct Connect service enables you to establish a private, physical network connection between AWS and your data center, office, or colocation environment. These private connections can provide a more consistent network experience than those made over the public internet.
For more information on the over 150 Direct Connect locations worldwide, visit the locations section of the Direct Connect product detail pages. Or, visit our getting started page to learn more about how to purchase and deploy Direct Connect.
Amazon CloudWatch now supports automatic enablement of Amazon CloudFront Standard access logs, AWS Security Hub CSPM finding logs, and Amazon Bedrock AgentCore memory and gateway logs and traces to CloudWatch Logs. Customers can set up enablement rules that automatically configure telemetry for both existing and newly created resources, ensuring consistent monitoring coverage without manual setup.
Enablement rules can be scoped to the organization, specific accounts, or specific resources based on resource tags to standardize telemetry collection. For example, a central security team can create a single rule to automatically send CloudFront access logs and Security Hub findings for all resources across their organization to CloudWatch Logs.
CloudWatch’s auto-enablement capability is available in all AWS commercial regions. Log ingestion will be billed according to CloudWatch Pricing.
Amazon CloudFront access logs and AWS Security Hub CSPM findings support organization-wide enablement rules. Bedrock AgentCore memory and gateway telemetry support account-level enablement rules. To learn more about enablement rules in Amazon CloudWatch, visit the Amazon CloudWatch documentation.
Amazon CloudWatch now supports automatic enablement of Amazon CloudFront Standard access logs, AWS Security Hub CSPM finding logs, and Amazon Bedrock AgentCore memory and gateway logs and traces to CloudWatch Logs. Customers can set up enablement rules that automatically configure telemetry for both existing and newly created resources, ensuring consistent monitoring coverage without manual setup.
Enablement rules can be scoped to the organization, specific accounts, or specific resources based on resource tags to standardize telemetry collection. For example, a central security team can create a single rule to automatically send CloudFront access logs and Security Hub findings for all resources across their organization to CloudWatch Logs.
CloudWatch’s auto-enablement capability is available in all AWS commercial regions. Log ingestion will be billed according to CloudWatch Pricing.
Amazon CloudFront access logs and AWS Security Hub CSPM findings support organization-wide enablement rules. Bedrock AgentCore memory and gateway telemetry support account-level enablement rules. To learn more about enablement rules in Amazon CloudWatch, visit the Amazon CloudWatch documentation.
3 nuevos modelos MAI de clase mundial ya disponibles en Foundry
Por: Mustafa Suleyman.
Presentamos MAI-Transcribe-1, junto a MAI-Voice-1 y MAI-Image-2. Calidad de clase mundial a velocidades relámpago, ahora disponibles a los precios más competitivos.
MAI-Transcribe-1 ofrece transcripción de voz a texto de última generación en los 25 idiomas más utilizados1, según el estándar industrial de FLEURS.2 Diseñado para ofrecer calidad de clase mundial en entornos reales y desordenados, su velocidad de transcripción por lotes es 2,5 veces superior a la de Microsoft Azure Fast que ya ofrece. Además, es muy eficiente, lo que hace que MAI-Transcribe-1 no solo sea el más preciso, sino también ultrarrápido. Ahora está disponible en Foundry con el mejor precio de cualquier proveedor de nube grande.
MAI-Voice-1 es nuestro modelo de generación de voz de primera gama. Construido para generar un habla natural y realista, rica en matices, rango emocional y expresión que preserva la identidad del hablante incluso en contenido de formato largo.
Hoy añadimos la capacidad de crear su propia voz personalizada de manera segura y protegida en Microsoft Foundry con solo unos segundos de audio. MAI-Voice-1 puede transformar la facilidad con la que los desarrolladores pueden crear experiencias y agentes de voz, con alta calidad y alta velocidad.
El modelo puede generar 60 segundos de audio en solo un segundo, y un uso muy eficiente de la GPU ofrece esa calidad y velocidad de forma asequible. Oír es creer, así que vívanlo por ustedes mismos con Copilot Audio Expressions o Copilot Podcasts.
MAI-Image-2 tiene un rendimiento y velocidad turbo en generación de imágenes en Copilot tras debutar como una de las 3 familias de modelos más destacadas en la clasificación de Arena.ai. Los usuarios experimentan tiempos de generación al menos el doble de rápidos en Foundry y Copilot con una calidad similar, basándose en datos reales de tráfico de producción. También se llevaban a cabo despliegues faseados en Bing y PowerPoint.
MAI-Image-2 fue creado con fotógrafos, diseñadores y narradores visuales que exigen luz natural, tonos y texturas de piel precisos, y texto claro dentro de la imagen para diagramas, maquetaciones y gráficos. Una vez más, la velocidad y la calidad no tienen costes elevados: MAI-Image-2 se ofrece a un precio competitivo por desempeño.
Los clientes ya han comenzado a adoptar MAI-Image-2 para trabajos creativos. WPP, uno de los mayores grupos de marketing y comunicación del mundo, es uno de los primeros socios empresariales que construyen a gran escala con MAI-Image-2.
«MAI-Imagen-2 es un verdadero cambio de juego. Es una plataforma que no solo responde a la compleja sutileza de la dirección creativa, sino que respeta a profundidad la pura destreza que implica generar imágenes reales listas para la campaña», dijo Rob Reilly, director creativo global de WPP. «WPP tiene algunos de los mejores talentos creativos del mundo y MAI-Image-2 los hace aún mejores.»
Imágenes creadas por WPP con MAI-Image-2
Modelos MAI: Mejores, más rápidos y más baratos que nuestros competidores.
Desplegamos estos modelos de primera gama con rapidez, para alimentar nuestros propios productos de consumo y comerciales. Estamos entusiasmados de compartir las mejoras en calidad, velocidad y eficiencia con nuestros clientes de Microsoft Foundry con precios muy competitivos.
· MAI-Transcribe-1 empieza en 0,36 dólares la hora.
· MAI-Voz-1 empieza en 22 dólares por cada 1 millón de caracteres.
· MAI-Image-2 comienza en 5 dólares por cada 1 millón de tokens para entrada de texto y 33 dólares por 1 millón de tokens para la salida de imagen.
Disponible ya en Microsoft Foundry y MAI Playground.
A partir de hoy, cualquier desarrollador puede construir con modelos MAI, incluido MAI-Transcribe-1, a través de Microsoft Foundry. También pueden probarlos en el MAI Playground (solo en EE. UU.).
¿Les interesan los modelos MAI pero no tienen acceso a Foundry? Rellenen este formulario y nos pondremos en contacto.
Modelos diseñados para ser mejores desde dentro hacia fuera.
En Microsoft AI, construimos una IA Humanista. Tenemos una visión distinta al crear nuestros modelos de IA: poner a los humanos en el centro, optimizar la manera en que las personas se comunican en verdad, entrenar para un uso práctico. Pronto verán más modelos nuestros en Foundry y directo en los productos y experiencias de Microsoft.
En consonancia con nuestro compromiso con una IA segura y responsable, estos modelos MAI fueron desarrollados, probados y seleccionados de manera rigurosa en equipo rojo. A través de Microsoft Foundry, los desarrolladores disponen de barreras integradas, gobernanza y controles empresariales diseñados para apoyar un despliegue seguro y conforme a gran escala.
1. Los 25 principales idiomas por uso de productos de Microsoft.
2. De entre las 25 principales lenguas globales, MAI-Transcribe-1 ocupa el primer lugar por FLEURS en 11 idiomas principales. Gana contra Whisper-large-v3 en los 14 restantes y contra Gemini 3.1 Flash en 11 de esos 14.
Amazon CloudWatch now supports native OpenTelemetry (OTel) metrics in public preview, enabling you to send metrics directly using the OpenTelemetry Protocol (OTLP) without custom conversion logic or additional tooling. You can now combine your custom OpenTelemetry metrics with AWS vended metrics from over 70 services and query them using PromQL — no additional agents or code changes required.
With native OTel support, a team running microservices on Amazon EKS and on-premises servers can now send OTel metrics from both environments directly to CloudWatch. They can correlate application-level metrics like order processing latency from their on-premises services with EKS pod CPU utilization and Application Load Balancer request counts, then use PromQL to build unified dashboards and alarms that span their entire infrastructure. CloudWatch anomaly detection works with OTel metrics, automatically identifying unusual patterns without requiring you to set static thresholds. Query Studio, a new console experience for PromQL, lets you write queries, explore metrics, create alarms, and build dashboards directly in the CloudWatch console.
Native OpenTelemetry metrics support is available in public preview in US East (N. Virginia), US West (Oregon), Asia Pacific (Sydney), Asia Pacific (Singapore), and Europe (Ireland). There is no charge for OpenTelemetry metrics or querying during preview. To learn more, see the Amazon CloudWatch OpenTelemetry documentation.
Amazon CloudWatch now supports native OpenTelemetry (OTel) metrics in public preview, enabling you to send metrics directly using the OpenTelemetry Protocol (OTLP) without custom conversion logic or additional tooling. You can now combine your custom OpenTelemetry metrics with AWS vended metrics from over 70 services and query them using PromQL — no additional agents or code changes required. With native OTel support, a team running microservices on Amazon EKS and on-premises servers can now send OTel metrics from both environments directly to CloudWatch. They can correlate application-level metrics like order processing latency from their on-premises services with EKS pod CPU utilization and Application Load Balancer request counts, then use PromQL to build unified dashboards and alarms that span their entire infrastructure. CloudWatch anomaly detection works with OTel metrics, automatically identifying unusual patterns without requiring you to set static thresholds. Query Studio, a new console experience for PromQL, lets you write queries, explore metrics, create alarms, and build dashboards directly in the CloudWatch console. Native OpenTelemetry metrics support is available in public preview in US East (N. Virginia), US West (Oregon), Asia Pacific (Sydney), Asia Pacific (Singapore), and Europe (Ireland). There is no charge for OpenTelemetry metrics or querying during preview. To learn more, see the Amazon CloudWatch OpenTelemetry documentation.
Amazon ECS announces Managed Daemons for ECS Managed Instances, enabling organizations to centrally deploy and manage software agents such as security, observability, and networking across their container infrastructure independent of application deployments. By decoupling daemon lifecycle management from application operations, Managed Daemons helps guarantee reliable agent coverage across all workloads, simplifies deployments and version updates, and improves resource utilization by running a single daemon task per managed instance.
With Managed Daemons, you can create a daemon for one or more Managed Instances capacity providers in your cluster. ECS places exactly one daemon task per managed instance and guarantees that daemons are running before any application tasks are placed, so cross-cutting functions such as logging, tracing, and metrics collection are always available. ECS orchestrates daemons as independent processes bound to the instance lifecycle rather than individual application tasks, allowing platform administrators to manage them independently from application teams. When you update daemon versions, ECS drains existing instances and provisions new instances with the updated daemon, automatically replacing service tasks with circuit breaker protection and rollback capabilities for reliable coverage across all your workloads.
To get started, you can use AWS Console, CLI, CloudFormation, or AWS SDKs to register a daemon task definition specifying your container image, then create a daemon with associated capacity providers in your clusters. This feature is now available in all AWS Regions. For more details, refer to our documentation and launch blog post. There is no additional cost – you pay only for the standard compute resources consumed by your daemon tasks.
Amazon ECS announces Managed Daemons for ECS Managed Instances, enabling organizations to centrally deploy and manage software agents such as security, observability, and networking across their container infrastructure independent of application deployments. By decoupling daemon lifecycle management from application operations, Managed Daemons helps guarantee reliable agent coverage across all workloads, simplifies deployments and version updates, and improves resource utilization by running a single daemon task per managed instance. With Managed Daemons, you can create a daemon for one or more Managed Instances capacity providers in your cluster. ECS places exactly one daemon task per managed instance and guarantees that daemons are running before any application tasks are placed, so cross-cutting functions such as logging, tracing, and metrics collection are always available. ECS orchestrates daemons as independent processes bound to the instance lifecycle rather than individual application tasks, allowing platform administrators to manage them independently from application teams. When you update daemon versions, ECS drains existing instances and provisions new instances with the updated daemon, automatically replacing service tasks with circuit breaker protection and rollback capabilities for reliable coverage across all your workloads. To get started, you can use AWS Console, CLI, CloudFormation, or AWS SDKs to register a daemon task definition specifying your container image, then create a daemon with associated capacity providers in your clusters. This feature is now available in all AWS Regions. For more details, refer to our documentation and launch blog post. There is no additional cost – you pay only for the standard compute resources consumed by your daemon tasks.
Amazon Bedrock is a fully managed service that provides access to a wide selection of high-performing foundation models from leading AI companies through a single API. Today, Amazon Bedrock expands structured outputs support to AWS GovCloud (US) Regions. Structured outputs enables foundation models to return consistent, schema-compliant, machine-readable responses—making it well-suited for government and regulated workloads that must meet strict compliance and data handling requirements.
Structured outputs helps with common production tasks, such as extracting key fields and powering workflows that use APIs or tools, where even minor formatting errors can break downstream systems. By ensuring schema compliance, it reduces the need for custom validation logic and lowers operational overhead by minimizing failed requests and retries—so you can confidently deploy AI applications that require predictable, machine-readable outputs. You can use structured outputs either by defining a JSON schema that describes your desired response format or by using strict tool definitions to ensure a model’s tool calls match your specifications.
Structured outputs is now generally available in all commercial AWS and AWS GovCloud (US) Regions where Amazon Bedrock is supported. To learn more about structured outputs and the supported models, visit the Amazon Bedrock documentation.
Amazon Bedrock is a fully managed service that provides access to a wide selection of high-performing foundation models from leading AI companies through a single API. Today, Amazon Bedrock expands structured outputs support to AWS GovCloud (US) Regions. Structured outputs enables foundation models to return consistent, schema-compliant, machine-readable responses—making it well-suited for government and regulated workloads that must meet strict compliance and data handling requirements.
Structured outputs helps with common production tasks, such as extracting key fields and powering workflows that use APIs or tools, where even minor formatting errors can break downstream systems. By ensuring schema compliance, it reduces the need for custom validation logic and lowers operational overhead by minimizing failed requests and retries—so you can confidently deploy AI applications that require predictable, machine-readable outputs. You can use structured outputs either by defining a JSON schema that describes your desired response format or by using strict tool definitions to ensure a model’s tool calls match your specifications.
Structured outputs is now generally available in all commercial AWS and AWS GovCloud (US) Regions where Amazon Bedrock is supported. To learn more about structured outputs and the supported models, visit the Amazon Bedrock documentation.
Amazon Simple Email Service (SES) Mail Manager now offers enhancements to email security and processing while simplifying email infrastructure migrations. These enhancements include optional TLS and certificate-based authentication (mTLS) support in Ingress Endpoint, and two new rule actions: Invoke Lambda function and Bounce.
These enhancements benefit organizations seeking to maintain compatibility with legacy systems while implementing stronger security controls, and advanced email routing capabilities. For example customers can now configure STARTTLS as an optional TLS configuration, enabling legacy systems that don’t support STARTTLS to connect to Mail Manager. With Mutual TLS (mTLS) in Ingress Endpoint customers can now used certificate-based authentication for enhanced security. The Invoke Lambda function rule action allows direct invocation of AWS Lambda functions from rule sets, enabling custom email processing workflows and the Bounce rule action provides RFC-compliant SMTP responses to sending servers.
These new enhancements are available today in all AWS Regions where Amazon SES Mail Manager is offered, except for the Middle East (UAE) and Middle East (Bahrain) regions. To learn more about Amazon SES Mail Manager and how these features can help streamline your email operations, visit https://aws.amazon.com/ses/.
Amazon Simple Email Service (SES) Mail Manager now offers enhancements to email security and processing while simplifying email infrastructure migrations. These enhancements include optional TLS and certificate-based authentication (mTLS) support in Ingress Endpoint, and two new rule actions: Invoke Lambda function and Bounce. These enhancements benefit organizations seeking to maintain compatibility with legacy systems while implementing stronger security controls, and advanced email routing capabilities. For example customers can now configure STARTTLS as an optional TLS configuration, enabling legacy systems that don’t support STARTTLS to connect to Mail Manager. With Mutual TLS (mTLS) in Ingress Endpoint customers can now used certificate-based authentication for enhanced security. The Invoke Lambda function rule action allows direct invocation of AWS Lambda functions from rule sets, enabling custom email processing workflows and the Bounce rule action provides RFC-compliant SMTP responses to sending servers.
These new enhancements are available today in all AWS Regions where Amazon SES Mail Manager is offered, except for the Middle East (UAE) and Middle East (Bahrain) regions. To learn more about Amazon SES Mail Manager and how these features can help streamline your email operations, visit https://aws.amazon.com/ses/.
