Current window
NVIDIA has made the full NVIDIA Run:ai platform available as open source software, giving customers, developers, and ecosystem partners access to production-grade AI workload and GPU orchestration capabilities. The release expands the NVIDIA DSX AI Factory Platform by establishing Run:ai as an open foundation for the broader ecosystem to build differentiated solutions.
SynthePulse InsightRead deep analysisThe NVIDIA KGMON team placed second in the KDD Cup 2026 Data Agents competition. Their system was built around the idea of making an agent's harness smaller, clearer, and easier to verify. The competition tasked agents with answering natural-language questions over heterogeneous data sources, including databases, CSV and JSON files, prose documents, PDFs, and briefing videos.
NVIDIA introduced DSX MaxLPS, a policy-governed power-sharing system designed to dynamically allocate power across AI factory resources. The technology aims to maximize throughput and efficiency by avoiding overprovisioning for rare peak-power moments. It enables customers to deploy more compute capacity within existing power budgets.
SynthePulse InsightRead deep analysisNVIDIA has introduced NV-Reason-CT, an open 3D CT vision-language model designed to support radiologist chain-of-thought reasoning. The model targets 3D CT scans, a modality underserved by current VLMs, aiming to improve volumetric imaging analysis. This addresses the gap where general-purpose and open medical AI models perform poorly on 3D CT data.
SynthePulse InsightRead deep analysisNVIDIA announced DLSS 5, introducing 3D-Guided Neural Rendering and granular controls that help game developers add lifelike lighting and material detail while preserving artistic intent. The update also includes NVIDIA ACE improvements, RTX Mega Geometry 2.0, and new RTX Kit capabilities across character AI, high-density geometry, and rendering workflows.
SynthePulse InsightRead deep analysisNVIDIA has introduced TensorRT Model Connect, an open collection of reference implementations designed to simplify deploying open AI models into native applications. The tool provides model-specific conversion, preprocessing, and post-processing code, enabling developers to deploy a model from checkpoint to inference in just two commands. This reduces deployment complexity and accelerates AI application development.
Alibaba has released the model weights for Qwen3.8-Flash-Next, a preview of the upcoming Qwen4 architecture. The model is a multimodal mixture-of-experts (MoE) with a 125B-parameter main model plus an additional 51B N-gram embeddings, activating 6B parameters per token. It features a native 262,144-token context window, extensible to 1M tokens, and is available for developers to experiment with and evaluate.
SynthePulse InsightRead deep analysisAlibaba has released the model weights for Qwen3.8-Flash-Next, a preview of the upcoming Qwen4 architecture, for developers to experiment with and evaluate. The model is a multimodal mixture-of-experts (MoE) model with 176B total parameters, activating 6B per token, and features a native 262,144-token context window extensible to 1M with YaRN.
SynthePulse InsightRead deep analysisNVIDIA has released Cosmos 3 Edge, a 4B-parameter edge AI model designed for robot control that runs on-device. It includes a 2B Nemotron-based reasoner and aims to help robots adapt to sensors, environments, and tasks.
Alibaba has released the open weights for Qwen3.8-2.4T-A95B (Qwen3.8-Max), its largest open-weight model, featuring 2.4 trillion total parameters with 95 billion activated per token. The model uses a fine-grained mixture-of-experts architecture with hybrid attention and supports configurable reasoning, deployable on NVIDIA GB300 NVL72, bringing near-frontier capabilities to the open ecosystem.
This article discusses the importance of choosing a full-stack observability solution for NVIDIA AI factories. It points out that AI infrastructure spans multiple layers, and when performance degrades, pinpointing the source can be difficult. A full-stack observability strategy connects telemetry across these layers, helping infrastructure and operations teams quickly detect and resolve issues.
NVIDIA has released JetPack 7.2.1, adding agentic video skills and T3000 emulation to the Jetson platform, aiming to enhance video processing and AI inference capabilities. The update targets applications like robotics and intelligent video analytics, improving the efficiency and flexibility of the video data pipeline.
NVIDIA has released Nemotron 3.5 Lightning, an open 30B mixture-of-experts model with 3B active parameters, designed for high-volume execution tasks in long-running AI agents, such as tool calls and result validation. The model aims to reduce latency and cost while maintaining accuracy.
NVIDIA has introduced NeMo Switchyard, a tool for routing AI agent workloads across multiple models to optimize cost, latency, and accuracy. It enables developers to dynamically select the most suitable model for each task, avoiding sending all requests to the largest model.
Meta has released Muse Glimmer, a 30B open-weight dense model with a 120K+ context window designed for local AI agentic workflows. Optimized for NVIDIA edge, desktop, and workstation platforms, it achieves 20K tokens per second on a single GPU, enabling always-on agents to process data locally and execute complex tasks.