Filters
- NVIDIA publishes AI Cluster Runtime packing Dynamo for Kubernetes
NVIDIA AI Cluster Runtime brings together NVIDIA Dynamo and required infrastructure with tested component versions and setup checks; the walkthrough covers configure, deploy, and validate on Kubernetes.
- Palantir says McCarthy fine-tuned Nemotron 3 Nano 4B for RFI classification
In its Q3 2026 business update, Palantir said McCarthy fine-tuned NVIDIA Nemotron 3 Nano 4B inside Palantir Foundry to classify requests for information on construction projects. Palantir’s chart shows about 73%–75% accuracy on that task; the update does not disclose the test sample size.
- NVIDIA open-sources Auto Ontology, a semantic layer that grounds agent SQL
NVIDIA released Auto Ontology under Apache 2.0: an open library that builds a semantic / ontology layer from enterprise metadata, reconciles definitions across systems, and exposes trusted metrics and lineage to agent harnesses through MCP, CLI, and API (with NL-to-SQL over connected warehouses). It complements existing semantic models rather than replacing them.
- NVIDIA RTX Spark Windows laptops add Nemotron 3.5 Transcribe, agent GPU sharing, and new local models
NVIDIA’s Local AI blog says RTX Spark Windows laptops from ASUS, Dell, HP, Lenovo, Microsoft and MSI gain open-weight Nemotron 3.5 Transcribe (promptable speech-to-text, speaker labels and translation across 40 languages on recordings), NVIDIA App GPU resource sharing so long-running agents can share the GPU with foreground apps, plus local support notes for Liquid AI LFM3, Nemotron 3.5 Bolt E4B, FLUX 3 Image Klein and DGX Spark 64GB availability.
- NVIDIA publishes Metatron-v1 humanoid robotics RL model on Hugging Face
NVIDIA published Metatron-v1 on Hugging Face, with a model card tagged reinforcement learning, robotics and humanoid under an nvidia-open-model license. Usage details are on the model card.
- NVIDIA publishes Metatron v1 G1 browser ONNX exports for Unitree G1 locomotion and recovery
NVIDIA’s Hugging Face card for nvidia/metatron-browser releases ONNX inference exports of the Metatron G1 walking/recovery policy used by the Metatron browser demo and G1 Racer, covering a 29-joint Unitree G1 policy with separate demo and batched racer graphs under an NVIDIA open-model license.
- NVIDIA and C-BRAIN deploy open-source AI tools to accelerate dementia and neurodegeneration research
NVIDIA’s blog says the Consortium for Biomedical Research and AI in Neurodegeneration (C-BRAIN) is working with NVIDIA to deploy open-source AI tools that help researchers synthesize large biomedical datasets for Alzheimer’s and other dementias, aiming to speed discovery where data volume outpaces expert review.
- NVIDIA publishes Nemotron-3.5-Parse, a ~0.63B document image parser with layout classes and boxes
NVIDIA's Hugging Face model card for Nemotron-3.5-Parse describes an open document-image parser that returns structured text with semantic layout classes, bounding boxes and reading order, combining a C-RADIOv4-SO400M vision encoder with a 10-layer mBART decoder (~0.631B unique trainable parameters). The card lists a Hugging Face release date of October 17, 2026, OpenMDW-1.1 licensing, Transformers and vLLM serving, and ParseBench / IndicVisionBench / MOSCAR evaluation numbers. DailyCited is adding the Official listing today after confirming the Hugging Face weights page.
- NVIDIA publishes InternVideo2-CLIP-anomaly-detection on Hugging Face
NVIDIA posts InternVideo2-CLIP-anomaly-detection on Hugging Face with TensorRT/ONNX/safetensors builds tagged for video-text retrieval, video embeddings, physical AI, and anomaly detection.
- NVIDIA unveils Minerva, an AIOps platform for end-to-end AI factory observability
NVIDIA launched Minerva, an AIOps platform that combines infrastructure and AI-workload telemetry with purpose-built models so teams can spot problems earlier and keep AI factories productive. NVIDIA positions it as end-to-end observability for complex AI factory operations.
- NVIDIA says Kazakhstan is expanding AI compute with planned Vera Rubin deployments and Firebird GPU capacity
NVIDIA’s blog says Kazakhstan’s AI stack is moving from plans to deployment, citing Firebird’s U.S. export license for 45,000 Grace Blackwell GPUs with Kazakhtelecom, plus planned NVIDIA Vera Rubin infrastructure and a wider local AI-apps layer.
- NVIDIA expands Cosmos with new world-model training tools, simulation checkpoints and Data Factory blueprints
NVIDIA says it has added new training tools, simulation checkpoints and Cosmos Data Factory blueprints to its Cosmos platform to help developers build, evaluate and deploy world models for physical AI.
- NVIDIA introduces Halos SAFE for autonomous-vehicle safety validation and Alpamayo 2 Edge for in-vehicle reasoning
At GTC Berlin, NVIDIA introduced Halos SAFE to speed up safety validation for autonomous vehicles and Alpamayo 2 Edge for reasoning inside the vehicle, as European carmakers showed Hyperion-based driving systems.
- Telecom operators add NVIDIA computing to data centers and cell sites, with AI-RAN pilots alongside Nokia
NVIDIA says leading telecom operators are adding its computing across data centers and cell sites to bring AI services closer to users, alongside live-network AI-RAN pilots with Nokia, announced at GTC Berlin.
- NVIDIA says it is expanding supply with Amkor, IBIDEN, KYEC, Samsung Electro-Mechanics, SPIL and Unimicron
NVIDIA says demand for compute is extraordinary and it is working with Amkor, IBIDEN, KYEC, Samsung Electro-Mechanics, SPIL and Unimicron to expand supply, from components to advanced packaging and testing.
- NVIDIA says Kumo Relational beats supervised machine learning on relational data with no training required
In a Hugging Face blog post, NVIDIA and Kumo researchers say the Kumo Relational model outperforms supervised machine learning on relational data without any task-specific training.
- Mistral and NVIDIA team up on AI for industrial engineering, with open benchmarks for physics and engineering agents
Mistral says it is working with NVIDIA on AI-native engineering workflows that pair Mistral Large 4 and Mistral's data generation and physics modeling pipelines with NVIDIA CUDA-X libraries and NVIDIA Warp, aiming to speed up engineering simulation. The companies will build public benchmarks and open datasets for evaluating physics AI models, and dedicated evaluations for AI agents in computer-aided engineering (CAE) and electronic design automation (EDA) that test tool use, multistep workflows, correctness, execution time and compute cost. Mistral will also integrate its models with NVIDIA Omniverse libraries for 3D scene generation and SimReady asset creation.
- NVIDIA cuDSS, a CUDA-X library for GPU-accelerated engineering and scientific simulation, is generally available
NVIDIA says cuDSS is now generally available. The CUDA-X library moves a critical part of engineering and scientific simulations onto GPUs, aimed at simulation-heavy design work from chips to aircraft and industrial systems.
- NVIDIA expands Confidential Computing so frontier model builders can protect models and data across global infrastructure
NVIDIA says it is expanding NVIDIA Confidential Computing, the infrastructure layer that protects models and data while they run, so frontier model builders can scale securely across global infrastructure. It ties the move to its new OpenShell Reference Architecture for deploying agents safely.
- NVIDIA says its cloud partners are expanding sovereign AI compute across Europe, the Nordics, Africa and Central Asia
At GTC Berlin, NVIDIA said its Cloud Partners in Europe, the Nordics, Africa and Central Asia are expanding regional compute capacity so developers can run open models close to their data on infrastructure built to be secure for AI agents.