Official updates
Announcements from official sources, ordered by publication time.
24 updates · Last 30 days
October 19, 2026 · UTC
Structured outputs for Anthropic Claude models on Amazon Bedrock
Amazon Web Services · Official · API & tools · Published
AI summary: Amazon Bedrock now offers schema-constrained JSON outputs and strict tool-parameter validation for Anthropic Claude models on bedrock-runtime and bedrock-mantle, using constrained decoding.
Developers can now access SageMaker Studio through the new AWS experience
Amazon Web Services · Official · API & tools · Published
AI summary: Amazon SageMaker Studio now lets new AWS customers sign up with existing social accounts or email and enter a preconfigured environment in about 20 seconds.
Amazon SageMaker JumpStart: 2026 year-to-date in review
Amazon Web Services · Official · API & tools · Published
AI summary: Amazon SageMaker JumpStart’s 2026 review reports more than 100 models added from 30 providers and new discovery and deployment features intended to simplify the path from model selection to production.
Pegasus 1.5 from TwelveLabs is now available in Amazon Bedrock
Amazon Web Services · Official · API & tools · Published
AI summary: TwelveLabs Pegasus 1.5 is available in Amazon Bedrock, generating timestamped video segments and structured metadata from user-defined descriptions without pre-indexing.
How Deepgram optimized voice AI at scale with AWS Trainium
Amazon Web Services · Official · API & tools · Published
AI summary: Deepgram and AWS ported the Nova-3 speech-to-text model to Trainium2 and optimized batch inference, achieving 11–70% more audio per dollar than Deepgram’s previous compute options.
Push-based KV transfer to improve TTFT and TPOT for disaggregated inference in vLLM
Amazon Web Services · Official · API & tools · Published
AI summary: AWS describes a push-based NIXL connector for vLLM disaggregated inference that overlaps KV destination registration with prefill and moves transfer setup off the critical path, improving TTFT and TPOT.
Improve TTFT for multi-turn agentic workloads with bi-directional KV transfer in vLLM
Amazon Web Services · Official · API & tools · Published
AI summary: AWS describes bi-directional KV transfer in vLLM, allowing prefill nodes to reuse decode-held KV blocks across agentic conversation turns and reducing TTFT as context grows.
October 16, 2026 · UTC
Building trust in AI agents: A production-ready evaluation framework with Amazon Bedrock AgentCore
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS post presents a nine-stage continuous evaluation framework using Amazon Bedrock AgentCore to keep production multi-agent systems safe, accurate, observable, and improvable.
Introducing Reflection Beam: an open-weight model for agentic coding on AWS
Amazon Web Services · Official · Models · Published
AI summary: This AWS post introduces Reflection Beam, an open-weight model for agentic coding, and documents deployment on SageMaker JumpStart or HyperPod Inference with JumpStart, SGLang, or vLLM.
October 15, 2026 · UTC
Meta’s Muse Spark Open now available in Amazon SageMaker JumpStart
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS post introduces Meta’s Muse Spark Open in Amazon SageMaker JumpStart and explains deployment, inference, infrastructure choices, and cleanup for its two variants.
Trust in Depth: Giving agents more autonomy without losing control
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS article introduces Trust in Depth, a layered approach for granting agents autonomy while retaining verifiable boundaries, identity, policy enforcement, and observability.
October 14, 2026 · UTC
Build self-improving document processing using Amazon Bedrock AgentCore
Amazon Web Services · Official · API & tools · Published
AI summary: The post presents an eight-agent document-processing system on Amazon Bedrock AgentCore that learns extraction instructions from new formats and reuses them for later documents.
Generate presentations with Amazon Bedrock AgentCore
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS post presents an Amazon Bedrock AgentCore system that generates grounded PowerPoint presentations from enterprise documents in 70–90 seconds using RAG, specialized tools, and AI-generated images.
How Tamnoon built AI-powered cloud security remediation with Amazon Bedrock
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS case study describes how Tamnoon built Tami, an Amazon Bedrock-based agentic orchestrator that safely investigates, prioritizes, and remediates cloud security findings, reducing critical exposure by over 90 percent in 90 days.
Deploy and evaluate models with Amazon SageMaker JumpStart
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS post presents a reusable SageMaker JumpStart workflow for discovering, deploying, benchmarking, and deleting model endpoints to compare models using representative prompts.
October 13, 2026 · UTC
Building agents that transact with Anchor Browser and AgentCore payments
Amazon Web Services · Official · API & tools · Published
AI summary: AWS describes how Anchor Browser integrates with Amazon Bedrock AgentCore payments to let AI agents autonomously access x402-protected web pages and APIs within managed browser sessions.
Manage private pricing offers for Amazon Bedrock with managed entitlements
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS post presents a lightweight script and AWS CDK automation for activating Amazon Bedrock managed entitlements from AWS Marketplace Private Pricing Offers across multiple AWS accounts.
Natural language security queries at Orca Security using Amazon Bedrock
Amazon Web Services · Official · API & tools · Published
AI summary: AWS describes Orca Security’s Discovery AI, a nine-stage Amazon Bedrock pipeline that converts natural-language security questions into validated Sonar queries while reducing cost, tokens, and latency.
How ShipStation Global built an AI IT help desk on Amazon Bedrock AgentCore
Amazon Web Services · Official · API & tools · Published
AI summary: ShipStation Global built Deckhand, a Slack-based AI IT help desk on Amazon Bedrock AgentCore that resolved routine questions, automated Jira workflows, and achieved 97.5% channel coverage with 96% positive feedback.
Build long-horizon agents with Grok 4.7 on Amazon Bedrock
Amazon Web Services · Official · API & tools · Published
AI summary: The AWS post evaluates a Grok 4.7 tool-calling agent on three search-based tasks, finding that higher reasoning effort improves accuracy while substantially increasing output tokens and latency.
Mixture-of-Experts kernel engineering on Amazon SageMaker AI
Amazon Web Services · Official · API & tools · Published
AI summary: This AWS post describes fused Mixture-of-Experts kernel optimizations for vLLM and Amazon SageMaker AI, delivering up to 10% higher throughput and 9% faster time to first token.
October 12, 2026 · UTC
Accessing OpenAI GPT models on Amazon Bedrock with global cross-Region inference in Thailand, Malaysia, Singapore, Indonesia, and Taiwan
Amazon Web Services · Official · API & tools · Published
Learn how to run OpenAI GPT-6 Sol, Luna, and Astra on Amazon Bedrock through global cross-Region inference from Singapore, Jakarta, Thailand, Malaysia, and Taipei. This post covers the Converse and OpenAI-compatible APIs, IAM and SCP permissions, quota sizing, and monitoring inference from your source Region.
Deploy decision models with llama.cpp on Amazon SageMaker AI
Amazon Web Services · Official · API & tools · Published
Learn how to deploy decision models with llama.cpp on Amazon SageMaker AI. This technical how-to serves OpenAI-compatible chat completions with the AWS Deep Learning Container for llama.cpp, then extends the runtime with SemIf to return probabilities over declared options on CPU or GPU.
Run multi-node inference workloads using Ray LLM Deep Learning Containers
Amazon Web Services · Official · Models · Published
Learn how to deploy one large language model across two GPU nodes on Amazon EKS using the multi-node Ray LLM Deep Learning Containers, with separate prefill and decode nodes behind one OpenAI-compatible endpoint. The pre-built container ships the full inference stack, so you configure the deployment instead of assembling it.