Train Qwen3 on AWS Trainium with torchtitan
A walkthrough published on the AWS Machine Learning blog shows how to train Qwen3 three ways on a single trn2.48xlarge instance using torchtitan-neuron release-2.33 (0.2.2+neuron.1), with torch and to…
A walkthrough published on the AWS Machine Learning blog shows how to train Qwen3 three ways on a single trn2.48xlarge instance using torchtitan-neuron release-2.33 (0.2.2+neuron.1), with torch and to…
AWS engineers contributed a push-based NIXL connector to the open-source vLLM inference engine that moves KV-cache transfer coordination off the critical path in prefill-decode disaggregated serving, …
I14 trained and served its Laplace 3B linear attention model on AWS Trainium2 using native PyTorch via TorchNeuron Native, reaching about 120,000 tokens per second in training on one trn2.48xlarge ins…
Smartsheet has launched a generally available Model Context Protocol (MCP) Server that lets AI assistants read, write and manage Smartsheet data through a standardized protocol spanning more than 40 t…
Smartsheet has launched a generally available Model Context Protocol (MCP) Server that lets AI assistants read, write and manage Smartsheet data through a standardized protocol spanning more than 40 t…
Amazon published a walkthrough showing how Amazon Quick, paired with Amazon Bedrock AgentCore as an intelligence layer exposing enterprise Model Context Protocol (MCP) servers, lets supply chain manag…
Fine-tuning GPT-OSS-20B with supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO) on Amazon SageMaker HyperPod raised the open-weight reasoning model's rate of reasoning in the r…
Octave built a multi-tenant AI foundation on Amazon Bedrock using Cohere Embed v4 that embedded 2 billion documents and cut deployment time per use case by 50 percent, the company reported in an AWS b…
Amazon Web Services made OpenAI's GPT-5.6 (Sol, Terra, Luna), GPT-6 (Astra, Sol, Luna) and GPT-6.1 Sol models available from the Canada (Central) and Canada West (Calgary) AWS Regions on Amazon Bedroc…
AWS released AWS Agent Registry, a managed service for cataloging and discovering Model Context Protocol (MCP) servers, agents and agent skills, so orchestrators can find tools at runtime instead of h…
AWS and Keyfactor published a joint reference architecture for giving AI agents cryptographic identity using X.509 public key infrastructure, built on Keyfactor's EJBCA Enterprise certificate authorit…
Eon, a cloud data protection platform founded by former AWS engineers, cut new-engineer onboarding to its multi-cloud investigation tooling from roughly two days to about two minutes by replacing per-…
AWS published a technical walkthrough showing how to build a semantic layer for agentic AI using the Denodo Platform and Amazon Bedrock AgentCore, federating Amazon Aurora, Amazon Redshift, and an on-…
Amazon Bedrock now supports structured outputs for all Anthropic Claude models on both the bedrock-runtime and bedrock-mantle endpoints, enforcing a supplied JSON schema during token generation throug…
Amazon Web Services announced that new Amazon SageMaker AI customers can sign up for AWS with an existing Amazon, Apple, GitHub, or Google account or an email address in 2–3 steps and land in a fully …
Deepgram ported its Nova-3 speech-to-text model to AWS Trainium2 and now runs eight model replicas on a single Amazon EC2 trn2.3xlarge instance, processing 11–70 percent more audio per dollar on offli…
Bi-directional KV cache transfer in vLLM cut time to first token by up to 1.6x for Qwen3-32B and up to 3.3x for zai-org/GLM-5-FP8 in multi-turn benchmarks on a 1P1D deployment of Amazon EC2 P5en insta…
Reflection AI released Beam, an open-weight sparse mixture-of-experts model for agentic coding and long-horizon software tasks, with 501 billion total parameters, 23 billion active parameters, and a r…
Meta's coding-optimized open-weight model Muse Spark Open is now available in Amazon SageMaker JumpStart, Amazon Web Services announced. The mixture-of-experts reasoning model has 816 billion total pa…
Amazon S3, including S3 Tables and S3 Vectors, is now natively available on the Vercel Marketplace and v0 by Vercel, letting developers provision storage resources without leaving Vercel. The integrat…