Amazon SageMaker

2026-10-29

CODING · 1 SOURCE · AWS Machine Learning

AWS Batch adds quota management to share GPUs for Amazon SageMaker Training jobs (SageMaker Python SDK v3)

AWS Batch introduced a quota management scheduling capability for Amazon SageMaker Training jobs that lets administrators define quota shares per team or project and configure resource-sharing strategies (RESERVE, LEND, LEND_AND_BORROW). The feature supports cross-share preemption to reclaim lent capacity and optional priority-based in-share preemption, and integrates with SageMaker via a Scheduling Policy, Service Environment, Job Queue, and examples in the SageMaker Python SDK v3.

6.0

2026-10-27

COMPANIES · 1 SOURCE · AWS Machine Learning

Amazon SageMaker adds SOCI-indexed images to speed AI training and processing jobs

Amazon announced support for Seekable OCI (SOCI) indexing in SageMaker AI Training and AI Processing jobs. SOCI enables lazy-loading of container images stored in Amazon ECR (you can use pre-indexed AWS Deep Learning Containers or add an index to your own image); SageMaker fetches only files needed to start the container and streams the rest, producing seconds-long startups in AWS benchmarks (measured pull speedups of about 8x–46x and ~34x on a 13.2 GB image across instance types) and up to a reported 90% reduction in GPU idle time, with no changes required to job definitions and automatic fallback to standard pulls.

7.0

2026-10-23

CODING · 1 SOURCE · AWS Machine Learning

CDK construct orchestrates SageMaker endpoint auto scaling during updates

AWS describes a reusable AWS CDK construct that replaces native Application Auto Scaling resources with three CloudFormation custom resources to deregister and then re-register auto scaling around SageMaker real-time endpoint updates. The approach (Deregister, Endpoint update, Register Scalable Target, Attach Scaling Policy) uses asynchronous custom-resource handlers, a ConfigHash to force re-invocation, and provider-generated Lambda and Step Functions helpers so endpoint configuration and auto scaling parameters can be updated in a single deployment without manual steps or broken CloudFormation stacks.

6.0

2026-10-22

COMPANIES · 1 SOURCE · AWS Machine Learning

Amazon SageMaker AI adds training session APIs to run custom training loops on managed GPUs

Amazon has launched SageMaker training session APIs that let developers write custom training loops (SFT, RL, preference optimization, etc.) while SageMaker handles the GPU compute, distribution, checkpointing, and fault recovery. The feature includes three SDK clients in the SageMaker Python SDK, an OpenAI-compatible sampling endpoint, and a published cookbook of SFT and RL recipes (including examples on GSM8K and LoRA workflows).

7.0

2026-10-21

MODELS · 1 SOURCE · Cohere

Cohere launches Rerank 5 Pro and Fast — passage-level reranking for enterprise retrieval

Cohere announced Rerank 5 (Pro and Fast), a new generation of reranking models that return passage-level relevance (Context Focus) to reduce LLM token costs and improve retrieval for agentic and enterprise workflows. Rerank 5 is generally available via the Cohere API, Model Vault, Microsoft Foundry, Amazon SageMaker and private deployments; Cohere says Pro leads public benchmarks (ViDoRe V3, nanoMTEB, BRIGHT, BEIR) and Fast offers lower-cost, lower-latency performance; API pricing is listed at $0.05/1M tokens (Pro) and $0.02/1M tokens (Fast).

7.0

2026-10-19

COMPANIES · 1 SOURCE · AWS Machine Learning

New AWS sign-in lets new users land directly in Amazon SageMaker Studio with $100 credits

AWS now lets new customers create an account using Google, GitHub, Apple, Amazon, or an email address and immediately land in a fully configured Amazon SageMaker Studio environment; eligible new accounts also receive $100 in Free Tier credits (plus a chance to earn another $100). The new sign-in is available only for new AWS accounts and includes prior Studio improvements—~20s environment creation, preconfigured permissions, Hugging Face deep links, and default access to G5/G6/G4dn GPUs for eligible accounts.

6.0

2026-10-19

MODELS · 1 SOURCE · AWS Machine Learning

Amazon SageMaker JumpStart expands catalog with 100+ models from 30 providers (Jan–Sep 2026)

Amazon SageMaker JumpStart, the model hub inside SageMaker, onboarded over 100 new models from 30 distinct providers between January and September 2026, adding contributions from vendors such as Alibaba (Qwen), NVIDIA (Cosmos3, Nemotron), Mistral, Google (Gemma 4), IBM (Granite), Cohere and others. JumpStart continues to offer model discovery, one‑click deployment, and console/SDK customization (LoRA, SFT, DPO, RLVR), and is accessible via SageMaker Studio, the Python SDK, and a Hugging Face deep-link integration.

7.0

2026-10-13

CODING · 1 SOURCE · AWS Machine Learning

Amazon SageMaker details fused MoE kernel optimizations for NVIDIA Hopper WGMMA constraints

Amazon SageMaker engineers describe kernel-level optimizations for fused Mixture-of-Experts (MoE) models, focusing on performance bottlenecks introduced by NVIDIA Hopper’s WGMMA instruction which fixes the M dimension at 64. The post explains how decode-time workloads (often M=1) and the large N dimensions of fused MoE (e.g., topk=8 aggregating many experts) create operand-shape asymmetry that starves Tensor Core lanes, and presents implemented optimizations and measured benefits using a representative MoE model.

7.0

2026-10-06

COMPANIES · 1 SOURCE · AWS Machine Learning

Manage Amazon SageMaker HyperPod Spaces directly from SageMaker Studio

AWS added UI support in SageMaker Studio to create and manage SageMaker Spaces on HyperPod EKS clusters, enabling data scientists to launch JupyterLab or Code Editor environments and start/stop Spaces without using the CLI. Administrators must install the SageMaker Spaces add-on on HyperPod clusters, attach required IAM policies, and enable per-user identity propagation for Studio domains.

6.0

2026-10-05

CODING · 1 SOURCE · AWS Machine Learning

Amazon SageMaker launches aws-ai-ml agent skill for inference optimization

Amazon introduced the aws-ai-ml skill available through the Agent Toolkit for AWS, which plugs into MCP‑compatible coding agents (e.g., Kiro, Claude Code, Codex) to benchmark SageMaker endpoints, recommend deployment configurations, compare performance runs, and generate executable SageMaker Python SDK v3 code. The skill can be used locally via the Agent Toolkit or inside Amazon SageMaker Studio JupyterLab and requires AWS credentials with SageMaker API permissions.

6.0

2026-09-25

MODELS · 1 SOURCE · AWS Machine Learning

Amazon SageMaker JumpStart enables real-time deployment of Qwen3-TTS-12Hz-1.7B-Base for voice cloning

Amazon documents how to deploy the publicly available Qwen3-TTS-12Hz-1.7B-Base text-to-speech model from SageMaker JumpStart to a fully managed, real-time inference endpoint using the SageMaker Python SDK; the model supports few-second reference-based voice cloning, cross-lingual synthesis across ten languages, streaming generation, and monitoring via CloudWatch. The post highlights benefits such as data control within an AWS account, cost alignment with compute usage, and use cases including localization, personalized assistants, and e-learning.

6.0

2026-09-22

STARTUPS · 1 SOURCE · The Decoder

Biological Computing's neuron-derived 'Neural Dynamics Adapter' to be offered via AWS

San Francisco startup The Biological Computing Co. (TBC) says it has a small proprietary software module — the Neural Dynamics Adapter — derived from measurements of cultured cortical neurons that can plug into open-source diffusion video models; TBC and Amazon Web Services plan to offer the adapter-backed text‑to‑video model on Trainium and via SageMaker/AWS Marketplace, with only an early-access signup available so far. According to TBC's press release and tests on the Oasis Minecraft model, the adapter (≈156k parameters, <0.1% of base size) reportedly preserves image detail longer and yields up to 5× faster generation and ~80% lower inference cost versus the unnamed base model, though TBC does not disclose the base model and the commercial offering is not yet public.

6.0

2026-09-18

COMPANIES · 1 SOURCE · AWS Machine Learning

Amazon SageMaker launches HyperPod Inference Gateway for GPU-aware LLM routing

Amazon announced the SageMaker HyperPod Inference Gateway, a Kubernetes-native EKS addon that routes OpenAI-compatible inference requests using real-time GPU signals (KV cache, queue depth, LoRA residency, etc.) to reduce first-token latency and GPU waste without application changes. The two-tier system offers per-cluster intelligent routing and fleet-wide coordination, deployable via a single InferenceGatewayConfig resource and emitting Prometheus/CloudWatch metrics.

7.0

2026-09-15

MODELS · 1 SOURCE · AWS Machine Learning

Walkthrough: Customize Qwen3-8B on SageMaker serverless to build an AI product-tagging system

AWS Machine Learning published a walkthrough that demonstrates customizing Qwen3-8B using supervised fine-tuning (SFT) and reinforcement learning with verifiable rewards (RLVR) via Amazon SageMaker serverless model customization, then deploying it for asynchronous inference to create a cost-efficient product tagging system.

5.0

2026-09-15

COMPANIES · 1 SOURCE · AWS Machine Learning

Amazon SageMaker AI adds instance preference lists for training and processing jobs

Amazon SageMaker AI now supports instance preference lists for training and processing jobs. You can specify an ordered list of up to five instance types, and SageMaker AI will automatically launch the first type with available capacity, removing the need for manual retry loops and capacity‑watching scripts.

6.0

2026-09-10

COMPANIES · 1 SOURCE · AWS Machine Learning

Amazon SageMaker HyperPod adds model caching to reduce inference cold starts

Amazon SageMaker HyperPod now supports model caching for inference: model weights and container images can be pre-loaded onto cluster nodes' local NVMe storage so pods read from local disk instead of downloading over the network. AWS says this cuts cold starts from tens of minutes to seconds; the announcement explains how it works and how to enable it.

7.0