All news with #amazon sagemaker ai tag

🟢 NVIDIA's Nemotron 3.5 Super VL is now available on Amazon SageMaker JumpStart, providing a multimodal vision-language foundation model that accepts images, video, and text and generates text. The model targets document intelligence, GUI understanding, video question answering, and multimodal agentic applications, supporting up to 1 million token context lengths and ten languages. SageMaker JumpStart offers optimized NVFP4 inference on G7 instances and BF16 reference weights on ml.p5en.48xlarge for customization.

🚀 Beam, an open weight model from Reflection, is now available in Amazon SageMaker JumpStart, enabling developers to build advanced coding agents and automated workflows while keeping data and infrastructure control. Beam runs alongside other SageMaker AI models and supports production scaling with built-in controls, audit trails, and monitoring. It has been ported and optimized for AWS Trainium3, offering a high-performance, cost-efficient deployment option.

🔐 Amazon SageMaker Studio now allows new users to sign up using Google, GitHub, or Apple credentials and land directly in a pre-configured Studio environment without a traditional AWS account setup. Users receive AWS Free Tier credits in most cases with no credit card required, can invite collaborators who sign in with their own credentials, and access shared models, notebooks, and endpoints without configuring IAM users. The feature is available in all AWS Commercial Regions where SageMaker Studio is supported.

🧭 SageMaker JumpStart now includes Muse Spark Open, Meta’s open-weight, coding-optimized foundation model tailored for agentic coding workflows. The model uses a Mixture-of-Experts architecture (816B total, 42B active) and supports a 1M-token context window for long-running projects and multimodal inputs. JumpStart provides BF16 and NVFP4 quantized variants, and customers can deploy endpoints in their own AWS accounts with a few clicks or via the SageMaker Python SDK.

🔑 Amazon SageMaker Studio now allows new users to sign up using Google, GitHub, or Apple credentials and land in a fully configured Studio environment without creating a traditional AWS account or providing a credit card. The change removes console navigation and IAM setup for first-time users, enabling immediate model fine-tuning, evaluation, and deployment. Teams can invite collaborators by email, who sign in with their own social credentials to access shared Studio resources without configuring IAM users or permissions.

🔒 Amazon SageMaker Studio now allows new users to sign up using Google, GitHub, or Apple credentials and land inside a fully configured Studio environment without creating a traditional AWS account or providing a credit card. New customers receive free credits under the AWS Free Tier and can immediately fine‑tune, evaluate, and deploy foundation models with preconfigured environments and permissions. Team members can be invited by email and sign in with their own social credentials to collaborate without manual IAM setup.

🛠️ The new AWS AI/ML skill for the Agent Toolkit for AWS lets coding agents select, train, optimize, and deploy models on Amazon SageMaker AI within a single natural-language workflow. The toolkit is now the supported integration point, replacing previous AWS Labs skills and plugins, and the skill is pre-installed in SageMaker Studio JupyterLab. Supported capabilities include inference optimization, end-to-end model customization and fine-tuning, serverful training, MLflow experiment tracking, JumpStart deployments, and processing jobs. The skill is available at no extra charge; you pay for the AWS resources consumed.

🛠️ Amazon SageMaker Unified Studio now supports custom Tooling blueprints, enabling administrators to define project foundations using their own AWS CloudFormation templates. Administrators can tailor environments to organization-specific requirements, including IAM role names and custom permissions boundaries. Templates are validated at registration and after deployment, support a wide range of CloudFormation resources, and work across accounts and Regions.

🛠️ Amazon Q Developer in the AWS Management Console now provides enhanced visualizations that reason across hundreds of AWS APIs to deliver comprehensive answers about an account. The update adds health dashboards with status indicators, interactive time-series charts with trend and anomaly detection, troubleshooting diagnostics, and detailed resource views. Responses are faster and more actionable, replacing multi-step text replies with immediate visual summaries. The features are available in all Regions where Amazon Q Developer is supported.

🔗 Amazon SageMaker Unified Studio now supports Iceberg REST Catalog (IRC) connections for external Apache Iceberg catalogs and IAM authentication for Amazon DocumentDB. IRC supports Snowflake Open Catalog, Databricks Unity Catalog, and a Generic IRC type, enabling browsing, ETL, and notebook access with OAuth2 or bearer tokens and short-lived S3 credentials. IAM authentication lets DocumentDB 5.0+ clusters with TLS validate connections via AWS STS without storing DB credentials. These capabilities are available now in all Regions where Unified Studio is offered at no extra cost.

🔥 Amazon EMR on EKS now supports interactive Apache Spark sessions via Spark Connect, enabling data engineers and scientists to develop and debug Spark applications from managed notebooks in Amazon SageMaker Unified Studio or their own IDEs like Jupyter and VS Code. An interactive session provides a persistent Spark context that spans cells and scripts, blending local Python execution with remote Spark operations on EKS. Sessions run as pods on virtual clusters secured by IAM execution roles and tagged by project and user, and Spark Connect is available in EMR release 7.14 (Spark 3.5) and emr-spark-8.1.0 (Spark 4.1) across AWS Commercial Regions.

🚀 Amazon SageMaker HyperPod Inference Gateway is a Kubernetes-native, GPU-aware routing add-on for EKS that integrates with existing SageMaker HyperPod infrastructure without application changes. It replaces round-robin balancing with real-time inference-signal-driven routing to reduce first-token latency by up to 82% and p99 TTFT by 97–98% in mixed-hardware and burst scenarios. The Gateway comprises an Envoy Endpoint, a Body-Based Router that reads model names from requests, and an Endpoint Picker that scores pods across six inference-level signals to select the optimal target. It supports OpenAI-compatible model servers like vLLM and SGLang, is available per-cluster in supported Regions, and will soon add cross-cluster, cross-region routing and centralized traffic controls.

🧭 Amazon announces general availability of CloudWatch Omni, an AI-driven observability platform organized around teams and applications. Omni integrates OpenTelemetry interoperability with CloudWatch scale, offering a standalone web console with SSO and local IDE extensions for hands-on agent work. It auto-discovers services, maps dependencies, and surfaces golden metrics across AWS accounts, Regions, and other clouds to streamline troubleshooting.

🔧 Amazon SageMaker AI now supports serverless model customization for the NVIDIA Nemotron 3.5 Lightning model, enabling supervised fine‑tuning (SFT), Direct Preference Optimization (DPO), and reinforcement fine‑tuning (RFT). The hybrid Mixture‑of‑Experts model has 3B active parameters and 30B total parameters. SageMaker handles infrastructure and orchestration so teams can focus on data and evaluation. Serverless customization is available in US East, US West, Asia Pacific (Tokyo), and Europe (Ireland).

🆕 Amazon SageMaker now supports instance preference lists for training and processing jobs, letting you submit prioritized sets of instance types and counts so SageMaker can pick the first available configuration. This reduces wait times and removes the need for complex retry logic or multiple concurrent submissions during high-demand GPU periods. You can include on-demand sources or reserved SageMaker Flexible Training Plans, and the feature is available today in all AWS Regions via CLIs, APIs, SDKs, and the Console.

🚀 Amazon SageMaker JumpStart now includes NVIDIA's Qwen3.6-35B-A3B-NVFP4 and Alibaba's Wan2.1-T2V-1.3B-Diffusers models, expanding available foundation models for AWS customers. Qwen3.6-35B-A3B-NVFP4 is a Mixture-of-Experts model optimized for agentic coding, multimodal and long-context reasoning, quantized to NVFP4 to reduce memory footprint while supporting very long context windows. Wan2.1-T2V-1.3B-Diffusers targets lightweight text-to-video generation, delivering 480p clips on consumer GPUs with modest VRAM requirements. Deployments are available via the SageMaker JumpStart model catalog or the SageMaker Python SDK.

🚀 Amazon SageMaker JumpStart now includes Ministral-3-3B-Instruct-2512 and Ministral-3-8B-Instruct-2512, two compact, vision-capable foundation models from Mistral AI designed for edge and resource-constrained deployments. These models offer multimodal understanding, multilingual instruction following, and native function calling, and can be deployed directly from the SageMaker JumpStart catalog or via the SageMaker Python SDK.

📣 Amazon SageMaker JumpStart now offers Google DeepMind’s Gemma-4-31B-it-assistant and NVIDIA-quantized Gemma-4-31B-IT-NVFP4, bringing the Gemma 4 31B dense architecture to enterprise workloads in full-precision and optimized 4-bit FP4 variants. The assistant-tuned model supports multimodal reasoning, large 256K-token contexts, and native function calling, while the NVFP4 variant reduces memory footprint and speeds inference for cost-efficient production. Deployments are available via the SageMaker console or Python SDK.

🆕 Amazon SageMaker JumpStart now includes three new foundation models: granite-speech-4.1-2b, kanana-2-30b-a3b-instruct, and OpenFold3. These models span multilingual ASR and speech translation, bilingual Korean–English instruction-following and agentic workflows, and all-atom biomolecular complex structure prediction. Customers can deploy these models directly from the JumpStart catalog or via the SageMaker Python SDK to accelerate AI use cases on AWS.

🚀 Amazon SageMaker HyperPod now supports model caching to speed up LLM inference by pre-loading model weights and container images onto cluster nodes. This reduces cold-start delays for scale-out events, allowing pods to start in seconds rather than minutes. The feature includes a weights cache on local NVMe and an image cache that pre-pulls container images, with automatic fallback to the original source if a cache is unavailable. Model caching is GA across all SageMaker HyperPod regions and is enabled via the HyperPod Inference Operator.