2/17/2026, 12:00:00 AM ~ 2/18/2026, 12:00:00 AM (UTC)

Recent Announcements

Amazon Bedrock reinforcement fine-tuning adds support for open-weight models with OpenAI-compatible APIs

Amazon Bedrock now extends reinforcement fine-tuning (RFT) support to popular open-weight models, including OpenAI GPT-OSS and Qwen models, and introduces OpenAI-compatible fine-tuning APIs. These capabilities make it easier for developers to improve open-weight model accuracy without requiring deep machine learning expertise or large volumes of labeled data. Reinforcement fine-tuning in Amazon Bedrock automates the end-to-end customization workflow, allowing models to learn from feedback on multiple possible responses using a small set of prompts, rather than traditional large training datasets. Reinforcement fine-tuning enables customers to use smaller, faster, and more cost-effective model variants while maintaining high quality.\n Organizations often struggle to adapt foundation models to their unique business requirements, forcing tradeoffs between generic models with limited performance and complex, expensive customization pipelines that require specialized infrastructure and expertise. Amazon Bedrock removes this complexity by providing a fully managed, secure reinforcement fine-tuning experience. Customers define reward functions using verifiable rule-based graders or AI-based judges, including built-in templates for both objective tasks such as code generation and math reasoning, and subjective tasks such as instruction following or conversational quality. During training, customers can use AWS Lambda functions for custom grading logic, and access intermediate model checkpoints to evaluate, debug, and select the best-performing model, improving iteration speed and training efficiency. All proprietary data remains within AWS’s secure, governed environment throughout the customization process. Models supported at this launch are: qwen.qwen3-32b and openai.gpt-oss-20b. After fine-tuning completes, customers can immediately use the resulting fine tuned model for on-demand inference through Amazon Bedrock’s OpenAI-compatible APIs - Responses API and Chat Completions API, without any additional deployment steps. To learn more, see the Amazon Bedrock documentation.

Amazon Connect now includes agent time-off requests in draft schedules

Amazon Connect now includes agent time-off requests in draft schedules, making it easier for you to view why an agent was not scheduled on a particular day or part of the day. For example, when generating schedules for next month, you can see that an agent who typically works Monday to Friday wasn’t scheduled for the first week because they’re on leave without needing to check the published schedules or troubleshooting configuration as to why agent was not scheduled. This launch helps schedulers quickly identify coverage gaps and adjust schedules before publishing them to agents.\n This feature is available in all AWS Regions where Amazon Connect agent scheduling is available. To learn more about Amazon Connect agent scheduling, click here.

Amazon EC2 C8a instances now available in the Europe (Frankfurt) and Europe (Ireland) region

Starting today, the compute-optimized Amazon EC2 C8a instances are available in the Europe (Frankfurt) and Europe (Ireland) regions. C8a instances are powered by 5th Gen AMD EPYC processors (formerly code named Turin) with a maximum frequency of 4.5 GHz, delivering up to 30% higher performance and up to 19% better price-performance compared to C7a instances.\n C8a instances deliver 33% more memory bandwidth compared to C7a instances, making these instances ideal for latency sensitive workloads. Compared to Amazon EC2 C7a instances, they are up to 57% faster for GroovyJVM allowing better response times for Java-based applications. C8a instances offer 12 sizes including 2 bare metal sizes. This range of instance sizes allows customers to precisely match their workload requirements. C8a instances are built on AWS Nitro System and are ideal for high performance, compute-intensive workloads such as batch processing, distributed analytics, high performance computing (HPC), ad serving, highly-scalable multiplayer gaming, and video encoding. To get started, sign in to the AWS Management Console. Customers can purchase these instances via Savings Plans, On-Demand instances, and Spot instances. For more information visit the Amazon EC2 C8a instance page.

Amazon MSK now supports dual-stack (IPv4 and IPv6) connectivity for existing clusters

Amazon Managed Streaming for Apache Kafka (Amazon MSK) now supports dual-stack connectivity (IPv4 and IPv6) for existing MSK Provisioned and MSK Serverless clusters. This capability enables customers to connect to Amazon MSK using both IPv4 and IPv6 protocols, in addition to the existing IPv4-only option. It helps customers modernize applications for IPv6 environments while maintaining IPv4 compatibility, making it easier to meet compliance requirements and prepare for future network architectures.\n Amazon MSK is a fully managed service for Apache Kafka that makes it easier for customers to build and run applications that use Apache Kafka as a data store. Previously, MSK Provisioned and Serverless clusters exclusively utilized IPv4 addressing for all connectivity options. With this new capability, customers can now enable dual-stack connectivity (IPv4 and IPv6) on existing MSK clusters using Amazon MSK Console, AWS CLI, SDK, or CloudFormation by modifying the Network Type parameter for a cluster from IPv4 to dual-stack. Upon successful update, MSK provisions IPv6-enabled network interfaces while maintaining existing IPv4 connectivity, ensuring uninterrupted service. To retrieve new IPv6 bootstrap broker strings for MSK Provisioned clusters, customers can use the GetBootstrapBrokers API to obtain the necessary connection information. All MSK Provisioned and Serverless clusters will retain IPv4-only connectivity unless explicitly updated. Dual-stack connectivity for existing MSK Provisioned and Serverless clusters is now available in all AWS Regions where Amazon MSK is available, at no additional cost. To learn more about Amazon MSK dual-stack support, refer to the Amazon MSK developer guide.

Claude Sonnet 4.6 now available in Amazon Bedrock

Starting today, Amazon Bedrock supports Claude Sonnet 4.6, which offers frontier performance across coding, agents, and professional work at scale. According to Anthropic, Claude Sonnet 4.6 is their best computer use model yet, allowing organizations to deploy browser-based automation across business tools with near-human reliability. Claude Sonnet 4.6 approaches Opus 4.6 intelligence at a lower cost. It enables faster, high-quality task completion, making it ideal for high-volume coding and knowledge work use cases. \n  

Claude Sonnet 4.6 serves as a direct upgrade to Sonnet 4.5 across use cases that require consistent conversational quality and efficient multi-step orchestration. For search and chat applications, it delivers reliable performance across single and multi-turn exchanges at a price point that makes high-volume deployment practical, maintaining quality standards while optimizing for scale. Developers can leverage Claude Sonnet 4.6’s for agentic workflows, seamlessly filling both lead agent and subagent roles in multi-model pipelines with precise workflow management and context compaction capabilities. Enterprise teams can use Claude Sonnet 4.6 to power domain-specific applications with professional precision, including spreadsheet and financial model creation that accelerates analysis workflows, compliance review processes that require meticulous attention to detail, and data summarization tasks where iteration speed and accuracy are paramount. Claude Sonnet 4.6 requires only minor prompting adjustments from Sonnet 4.5, ensuring smooth migration for existing implementations. 

 

Claude Sonnet 4.6 is now available in Amazon Bedrock. For the full list of available regions, refer to the documentation. To learn more and get started with Claude Sonnet 4.6 in Amazon Bedrock, read the About Amazon blog and visit the Amazon Bedrock console.

AWS Blogs

AWS Japan Blog (Japanese)

AWS Big Data Blog

AWS Database Blog

AWS Developer Tools Blog

Networking & Content Delivery

AWS Storage Blog

Open Source Project

AWS CLI

Amplify for JavaScript

Amplify for iOS

Bottlerocket OS