Concise summaries of official AWS update announcements for Amazon Bedrock.
Information as of November 3, 2026.
Amazon Bedrock now offers OpenAI's Daybreak Red in AWS GovCloud (US-East), enabling authorized security professionals to use advanced AI for vulnerability research, exploit reproduction, and mitigation development with enhanced controls and privacy
Amazon Bedrock now supports OpenAI GPT-5.6 models (Terra and Luna) on the bedrock-runtime endpoint and introduces Cross-Region Inference in AWS GovCloud (US), enabling customers to achieve higher throughput without managing capacity across multiple regions.
AWS Step Functions workflows can now be used as tools for AI agents via Amazon Bedrock AgentCore Gateway, allowing agents to decide when to run workflows and what input to pass.
AWS Agent Registry, a fully managed discovery service via Amazon Bedrock AgentCore, is now available in four additional regions: Europe (London), Europe (Frankfurt), US East (Ohio), and Canada (Central).
Amazon Bedrock now provides cross-Region inference in Canada, allowing customers with strict data residency requirements to keep inference processing entirely within Canada
Amazon Bedrock's Web Search now supports domain filtering and user location controls, enhancing response precision. These are configured directly in existing OpenAI Responses API requests.
Amazon Bedrock AgentCore Gateway now supports InvokeModel APIs and model fallback, enabling access to multiple model providers through a single governed endpoint without client code changes
Amazon Bedrock now supports structured outputs for all Anthropic Claude models across both the bedrock-runtime and bedrock-mantle endpoints, enabling schema-compliant JSON responses without custom parsing or post-processing.
Amazon Bedrock Advanced Prompt Optimization now supports prompt shortening, an optional setting that reduces prompt length while maintaining or improving evaluation scores. Optimized prompts are returned only when shorter and score at least as high as the original.
Amazon Bedrock now supports the reasoning.summary parameter for OpenAI models, allowing developers to retrieve a human-readable summary of a model's reasoning alongside its response
Amazon Bedrock now supports the TwelveLabs Pegasus 1.5 model, which generates structured, time-coded metadata from video, allowing applications to query specific moments without prior indexing.
Amazon Bedrock now offers Claude Haiku 5.5 on AWS GovCloud (US), the fastest and most efficient model in the Claude 5.5 family, designed for subagents and high-volume, cost-sensitive workloads.
AWS now offers Anthropic's fastest and most efficient model, Claude Haiku 5.5, for subagents and high-volume, cost-sensitive tasks, reducing costs by approximately 75% compared to Claude Haiku 4.5.
Amazon Bedrock now supports Z.ai's GLM 5.3, providing a powerful option for agentic coding and long-horizon software engineering work with the latest open-weight model and AWS security and compliance.
AgentCore Gateway now supports TLS certificates signed by private certificate authorities on MCP, OpenAPI, and HTTP proxy targets, enabling secure connections to targets using TLS certificates from your own private CA.
Amazon Bedrock now supports in-region inference for Anthropic's Claude Opus 5.5 and Claude Sonnet 5 in the UK (London) Region, enabling customers in financial services, healthcare, and the public sector to process data at scale within the country
Amazon Bedrock now offers UltraFast mode for OpenAI's GPT-6 Astra, delivering up to 6x faster API inference at up to 300 tokens per second for latency-sensitive applications.
Amazon Bedrock now offers Anthropic's Claude Opus 5, Claude Sonnet 5, and Claude Haiku 4.5 in India, Opus 5 and Sonnet 5 in South Korea, and Sonnet 5 in Singapore, enabling customers to process data locally at scale
Amazon Bedrock Managed Agents, developed jointly by AWS and OpenAI, uses a customized version of OpenAI's Agents API integrated with AWS resources. No additional charges apply during the preview period.
Amazon Bedrock now offers Claude Sonnet 5.5 on AWS GovCloud (US), enhancing efficiency and accuracy for coding and knowledge work at lower cost
AWS now offers Claude Sonnet 5.5, a smarter and more efficient model that delivers improved performance for coding and knowledge work at lower cost and faster speed.
Amazon Bedrock Managed Knowledge Base now offers native connectors for Salesforce and Zendesk, enabling customers to automatically sync content from these platforms
Amazon Bedrock now generally available with OpenAI's GPT-6 Sol and GPT-6 Luna. Sol is for complex tasks and software development, while Luna is for high-volume tasks, both supporting maximum context tokens.
AWS now offers Anthropic's latest model, Claude Opus 5.5, which efficiently handles long-running coding and knowledge work and provides clear reports on actions taken and next steps. Available via Amazon Bedrock and Claude Platform on AWS.
Amazon Bedrock now offers Anthropic's latest model, Claude Opus 5.5, on AWS GovCloud (US), enhancing long-running coding and knowledge work with clear reporting and improved efficiency.
Amazon Bedrock AgentCore now offers a next-generation serverless microVM compute with improved memory management and faster cold starts.
Amazon Bedrock's Web Search is now available in AWS GovCloud (US-West), providing compliance-aware web results for government and public sector workloads
Amazon Bedrock AgentCore Identity provides a managed consent portal, eliminating the need for developers to build custom OAuth callback infrastructure
Amazon Bedrock now offers Anthropic's new Fable 5.1 model in AWS GovCloud (US), designed for long-running, high-stakes tasks and managing entire software projects
Anthropic's most capable frontier model now delivers improved intelligence for coding, research, and enterprise workflows
AWS Agent Registry now generally available, offering centralized management and discovery of agents and tools across organizations with support for CloudFormation, Terraform, and cross-account sharing
Amazon Bedrock AgentCore now supports two additional AWS regions, enabling users to deploy agents closer to their end users with reduced latency.
Amazon Bedrock in AWS GovCloud (US) now supports SpaceXAI's Grok 4.6, designed for coding and agentic tasks with a 500k context window
Amazon Bedrock now offers OpenAI's GPT-5.6 models, enabling advanced tasks such as autonomous coding, genomics analysis, and cybersecurity research with high performance and efficiency.
Amazon Bedrock lowers prices for OpenAI GPT-5.6 Sol by 20% for input and 33.3% for output, effective until at least November 21, 2026
Amazon Bedrock AgentCore now supports domain and published date filtering for web search, expanding to Europe and Asia Pacific regions, allowing per-request control over trusted sources and time windows.
AgentCore payments enables AI agents to autonomously discover, access, and pay for paid APIs, MCPs, and content with minimal code, providing security and observability for enterprises
Amazon Bedrock now offers OpenAI's Daybreak models for security teams
Amazon Bedrock now offers Claude Opus 5 in AWS GovCloud (US) regions, delivering advanced coding, long-running agents, and complex document analysis capabilities with zero data retention enabled by default
Amazon Nova Multimodal Embeddings, a unified model supporting text, documents, images, video, and audio, enables high-accuracy cross-modal search in AWS GovCloud (US-West)
Amazon Bedrock now allows IAM users and roles as cost allocation tags, enabling analysis of model inference costs through the bedrock-mantle endpoint by team or project
Amazon Bedrock AgentCore now offers memory for conversation context, centralized policy controls using Cedar language, and a managed harness allowing developers to build agents with minimal code in AWS GovCloud (US-West).
Amazon Bedrock AgentCore now offers runtime instances, enabling users to run agents on their own Amazon EC2 instances without managing infrastructure, providing scalable and secure deployment for AI agents
Amazon Bedrock AgentCore adds new controls for stateful authorization and traffic management. Temporal policies enforce workflow sequencing and data freshness, while rate limiting allows per-user or per-group traffic controls to tools, models, and agents.
Amazon Bedrock now offers Web Search, enabling OpenAI models to ground responses with current web knowledge entirely within AWS, eliminating third-party provider onboarding and additional security reviews.
Amazon Bedrock's OpenAI GPT-5.6 models now support 1 million token context windows, enabling processing of entire codebases and long documents in a single request
AWS launches an open-source Context Ontology Accelerator to help organizations build machine-readable business models, enabling AI agents to make consistent, explainable decisions through a knowledge graph and Model Context Protocol.
Amazon Bedrock reduces prices for OpenAI GPT-5.6 Luna and Terra models by up to 80%, enabling customers to apply their capabilities across more applications and process larger workloads
Amazon Bedrock introduces Google DeepMind's Gemma 4 models for generative AI applications in reasoning, multimodal understanding, and software engineering workflows
Amazon Bedrock adds xAI's Grok 4.3 model to AWS GovCloud (US-West), offering enhanced reasoning and agent-building capabilities ideal for enterprise workloads.
Offered via Amazon Bedrock and Claude Platform on AWS with zero data retention, enhancing code generation, long-running agents, and complex analysis accuracy
Amazon Bedrock offers Claude Sonnet 5 in AWS GovCloud (US), delivering strong performance across coding, agentic tasks, and knowledge work while keeping data within AWS infrastructure.
Amazon Bedrock AgentCore delivers agent traces and prompts to the same log group as agent logs, providing unified observability for AI agents in a single Amazon CloudWatch log group.
AWS Data Exports introduces consistent, structured attributes for Amazon Bedrock costs, eliminating the need for custom logic in CUR 2.0
Amazon Bedrock AgentCore, the platform for building, connecting, and optimizing agents, is now available in four new AWS Regions: Asia Pacific (Bangkok), Asia Pacific (Malaysia), Europe (Milan), and Europe (Spain).
Amazon Bedrock AgentCore has increased default runtime quota limits, allowing users to run more AI agents simultaneously
Available via Amazon Bedrock and Claude Platform on AWS, delivering high performance for coding, agents, and knowledge work
Amazon Bedrock now offers Claude Opus 4.8, enhancing capabilities for agentic coding, knowledge work, and long-running tasks in AWS GovCloud (US)
Amazon Bedrock now supports these models certified for FedRAMP High and DoD IL-4/5 in AWS GovCloud (US), enabling government agencies to build compliant generative AI applications with high-performance serverless inference powered by Mantle.
Amazon Bedrock Guardrails now offers new automated policy refinement workflows. The iterative policy improvement and ambiguity reduction workflows help customers enhance their policies with less manual effort, leading to more reliable Guardrail validation results.
Amazon Bedrock AgentCore Memory enables multi-account architectures, allowing memory resources and agents across AWS accounts. Use resource-based policies to grant access and configure delivery to separate accounts.
Amazon Bedrock AgentCore now generally available with managed agent harness, allowing users to define agents through configuration and switch models mid-session
Amazon Bedrock AgentCore now offers Web Search, allowing developers to build agents with up-to-date web knowledge without integrating external services
Amazon Bedrock Guardrails now offers automated reasoning checks using formal verification techniques to validate AI model outputs with mathematical rigor, significantly reducing risks in regulated industries and enterprise applications
Amazon Bedrock Guardrails now offers the InvokeGuardrailChecks API, allowing users to apply individual safeguards at any point in agentic AI applications without creating guardrail resources. The API provides granular control over which safeguards to run at each step, returning severity and confidence scores for custom thresholds and actions.
Amazon Bedrock Managed Knowledge Base, a fully managed retrieval-augmented generation service, is now generally available, allowing developers to build production-ready AI agents grounded in enterprise data without managing vector databases, data pipelines, or retrieval infrastructure
Amazon Bedrock AgentCore now supports Bedrock Guardrails in policy, providing enterprises with deeper safety and security controls for scaling AI agents in production by evaluating agent actions and gateway calls in real-time.
Amazon Bedrock AgentCore now offers optimization capabilities that turn production traces into continuous agent improvements, identifying and fixing silent failures that affect users.
Amazon Bedrock AgentCore Memory now allows attaching metadata directly from applications for long-term memory records, ensuring values pass through extraction unchanged without LLM inference.
Claude Fable 5 delivers Mythos-level capabilities combined with additional safeguards, and can be accessed through Amazon Bedrock or the Anthropic-operated Claude Platform on AWS. Claude Mythos 5, the same model without those safety classifiers, is available only to a small group of customers.
No updates match the selected filters. Try changing them.