NVIDIA

2026-10-01

MODELS · 1 SOURCE · Hugging Face

Olmo launches Olmo-core 3 — open, scalable MoE training infrastructure

Olmo has released Olmo-core 3, a redesigned open mixture-of-experts (MoE) training framework intended to scale MoE models into the trillion-parameter range while preserving computational efficiency. The stack replaces an FSDP-based implementation with a DDP-centered system, combines expert/pipeline parallelism and other optimizations, reports a 2.7× throughput uplift on a 47B-parameter MoE across eight NVIDIA B300 GPUs, and shows MXFP8 delivering about 21% higher throughput than BF16 in controlled tests.

7.0

2026-09-30

MODELS · 1 SOURCE · NVIDIA Developer

NVIDIA publishes Dynamo-Triton HSTU recsys-examples for production inference

NVIDIA's recsys-examples repository provides an end-to-end HSTU generative recommender inference workflow that combines Dynamo-Triton, PyTorch AOTI, FlexKV-backed KV caching, and NV Embedding Cache. On an RTX PRO 6000 Blackwell GPU, the workflow's KV-cache-backed AOTI deployment achieved up to 5.93x lower latency (batch size 8, 100% GPU KV-cache hit) versus the same AOTI configuration without caching, and the guide shows export, validation (Python and native C++), and serving steps.

7.0

2026-09-30

CODING · 1 SOURCE · NVIDIA Developer

NVIDIA makes cuObject client/server (Server 2.0.0) generally available and ships SCADA Server SDK

NVIDIA announced the general availability of cuObject client and server libraries (cuObject Server 2.0.0), providing standardized APIs and an RDMA wire protocol for accelerated, zero-copy object storage access from GPUs. NVIDIA also released a SCADA Server SDK to let storage providers build servers that respond to GPU-initiated SCADA requests, expanded the xio-sig consortium to include cuObject alongside cuFile, and said Microsoft plans to join the xio-sig board while Google Cloud is evaluating participation; IBM Storage demonstrated a prototype integrating SCADA with IBM Storage Scale.

8.0

2026-09-29

MONEY · 1 SOURCE · TechCrunch AI

OpenAI reportedly in talks to raise $30B at $1.4T valuation

Bloomberg reports OpenAI is in talks with investors to raise at least $30 billion in a pre-IPO funding round at an approximate $1.4 trillion valuation; the round would serve as a bridge to an eventual IPO. The report cites a 70% jump in run-rate revenue to $40 billion in August and notes OpenAI previously raised $122 billion in March at an $852 billion valuation; CEO Sam Altman has said OpenAI will not pursue an IPO in 2026 to prioritize AI safety.

8.0

2026-09-29

CODING · 1 SOURCE · NVIDIA Developer

NVIDIA publishes lessons from building AI-native TensorRT Model Connect

NVIDIA published a first‑hand account of engineering choices behind TensorRT Model Connect, an open‑source C++ collection of model reference implementations built on TensorRT. The article describes an AI‑native development approach—using coding agents with outcome‑driven tasks, model‑family isolation, reversible changes, and GPU‑backed validation—and notes the project covered 128 model families tested on NVIDIA GB300 as of the public July 29, 2026 release comparison.

6.0

2026-09-29

COMPANIES · 1 SOURCE · NVIDIA Developer

NVIDIA releases VSS Blueprint 3.3 with Build Vision Agent skill and Adaptive EVS

NVIDIA’s Metropolis Video Search and Summarization (VSS) Blueprint 3.3 adds the vss-build-vision-ai skill to compose multi-workflow visual AI deployments from a single prompt and introduces Adaptive Efficient Video Sampling (EVS) to prune redundant visual tokens and batch VLM work. NVIDIA cites a demo that deployed a bottling-line agent in under 30 minutes on an RTX PRO 6000 Blackwell and reports up to ~80% fewer VLM input tokens for a 60-minute summary and a 46% increase in concurrent real-time VLM streams in their tests; the repository and agent skills are available and a live demo is scheduled for Oct. 1 at 9 a.m. PT.

7.0

2026-09-29

COMPANIES · 3 SOURCES · TechCrunch AI · The Verge AI · The Decoder

AMD to acquire Fei‑Fei Li’s World Labs for $8.2 billion

AMD will acquire World Labs, a developer of 'world models' founded by Fei‑Fei Li, in an $8.2 billion deal; Li will join AMD as executive vice president and chief scientist. World Labs’ first product, Marble, and the team's focus on models that understand physical reality are cited as drivers for integrating model research with AMD’s chip roadmap; the deal is expected to close before year‑end pending regulatory approval.

9.0

2026-09-29

MODELS · 1 SOURCE · Hugging Face

NVIDIA releases Kumo Tabular, an open foundation model for tabular prediction

NVIDIA released Kumo Tabular, an open-source foundation model for tabular classification and regression available on Hugging Face and GitHub under the OpenMDW-1.1 license. The Transformer-based model family (28M–215M parameters), pretrained on synthetic tables, predicts labels in a single forward pass with no task-specific training and tops benchmarks TabArena, BeyondArena, TALENT and ScoringBench.

8.0

2026-09-28

COMPANIES · 1 SOURCE · NVIDIA Developer

NVIDIA publishes OpenShell and Open Agent Safety Platform for continuous in-silicon agent monitoring

NVIDIA introduced OpenShell, an Apache-2.0 open-source secure runtime for sandboxed autonomous AI agents, and described the NVIDIA Open Agent Safety Platform that pairs OpenShell on Vera CPUs with NVIDIA Sentry on BlueField-4 DPUs to provide layered, out-of-band monitoring and real-time policy enforcement. The platform emphasizes verifiable policies, out-of-band enforcement, controlling the path to the model, scaling agent authority with reasoning visibility, and a shared responsibility model across labs, enterprises, and hardware providers.

8.0

2026-09-28

CODING · 1 SOURCE · NVIDIA Developer

NVIDIA releases OpenShell 0.1.0 to enforce runtime controls for AI agents

NVIDIA published OpenShell 0.1.0, an open-source runtime that enforces which systems, data, and services AI agents can access without rewriting the agents. The release includes Gateway, Supervisor, and Sandbox components, a formal policy prover, credential protection, and support for deployment via Docker and Kubernetes; organizations such as Cadence, Slack, and Gecko Robotics are already adopting it.

7.0

2026-09-28

COMPANIES · 1 SOURCE · NVIDIA Developer

NVIDIA DSX MaxLPS boosts usable GPUs and AI throughput via policy-governed power sharing

In a joint evaluation with Nscale, NVIDIA’s DSX MaxLPS used policy-governed power sharing on GB300 NVL72 systems at Nscale’s Verne campus to run Kimi K2.5 workloads, allowing 192 GPUs under the same 264.4 kW provisioned power budget versus a 140-GPU static baseline and increasing normalized aggregate throughput by 49.2% (tokens/s/W from 4.10 to 6.12). Median and P75 latency remained within 5% of baseline while P99 time-to-first-token rose 17%; NVIDIA outlines a five-stage validation process operators should follow before scaling deployment.

6.0

2026-09-24

STARTUPS · 1 SOURCE · TechCrunch AI

Lovable reports $600M annual run rate as its vibe-coding platform grows

Lovable’s co-founder Fabian Hedin said at the HumanX summit that the company has crossed an annualized run-rate revenue of $600 million, up from about $500 million in June. Hedin said the startup is focusing on enterprise customers (he claimed two-thirds of Fortune 500 use the product), its apps attract nearly a billion visits per month, and the company has raised over $700 million in two rounds in the past eight months, most recently $400 million in August at a $13.3 billion valuation.

7.0

2026-09-23

MODELS · 1 SOURCE · NVIDIA Developer

NVIDIA releases NV-Reason-CT — open 3D CT VLM with radiologist-style chain-of-thought

NVIDIA introduced NV-Reason-CT, an open research 3D vision-language model that combines a full 3D vision transformer encoder with a Qwen3.5-4B language model trained to generate radiologist-style chain-of-thought reasoning and structured reports for chest and abdominal CT. The model reports state-of-the-art results on the CT-RATE benchmark (Macro-F1 0.614, Macro-AUROC 0.871), was judged clinically plausible by NIH radiologists, and is offered as a foundation for post-training and integration with NVIDIA Medical AI tools (segmentation, synthetic data).

8.0

2026-09-23

CODING · 1 SOURCE · NVIDIA Developer

NVIDIA open-sources NVCRE, a Kubernetes controller to validate GPU cluster readiness

NVIDIA Cluster Readiness Engine (NVCRE) is an open-source Kubernetes controller that runs real distributed workloads across topology-aware node groups to validate GPU cluster readiness before production AI jobs start. It exposes Certification/Workflow/Job CRDs, adaptive fault isolation to pinpoint failing nodes, a WorkloadRun API for multi-node GPU setup, integrates with NVIDIA AI Cluster Runtime and NVSentinel (part of NVIDIA DSX OS), and is installable via nvcrectl; contributors can extend tests and workload adapters on GitHub.

7.0

2026-09-23

CODING · 1 SOURCE · Hugging Face

NVIDIA demonstrates using Warp and MJWarp to run MuJoCo at GPU scale (up to 2,048 envs)

NVIDIA describes migrating a MuJoCo workflow (an SO-101 follower arm) to MJWarp, an implementation of MuJoCo’s physics pipeline in NVIDIA Warp, and demonstrates advancing as many as 2,048 parallel environments on NVIDIA GPUs. The article explains Warp’s GPU kernel model, device arrays, composable launches and graph capture, and notes Warp features such as differentiable kernels and an opt-in deterministic mode (Warp ≥ 1.15); it focuses on preparing and scaling simulation environments rather than training policies.

6.0

2026-09-23

COMPANIES · 1 SOURCE · NVIDIA Developer

NVIDIA open-sources NodeWright, a Kubernetes-native host package manager for GPU fleets

NodeWright is an open-source, Kubernetes-native package manager that declaratively configures and updates host operating systems across GPU clusters while minimizing disruption to running workloads. It integrates with NVIDIA DSX OS components — including NVIDIA AI Cluster Runtime, NVCRE, and NVSentinel — supports progressive rollout policies, and is installable via Helm.

6.0

2026-09-23

MODELS · 1 SOURCE · Hugging Face

NVIDIA releases Nemotron 3 Diarization — open-weight 100M model for real-time multi-speaker diarization

NVIDIA published Nemotron 3 Diarization, an open-weight, 100M-parameter speaker-diarization model that ranks #1 on Voice Arena's Diarization-Bench (14.72% DER). The model supports up to eight anonymous speaker channels, handles overlapping speech in streaming and offline modes, and uses arrival-order speaker caching (AOSC) and a FIFO context buffer; training included public and licensed data, with David AI data reducing compound DER by 0.77 points.

7.0

2026-09-22

CODING · 1 SOURCE · NVIDIA Developer

NVIDIA documents TensorRT LLM adaptations for confidential Blackwell (B200) inference

NVIDIA published guidance and measured results showing how TensorRT LLM adapts to confidential computing on Blackwell B200 hardware, including CC-aware memory selection, async token readback, using the GPU %globaltimer for autotuning, and communication-path detection when NVLS multicast is unavailable. In their tests, CC-enabled runs retained about 96.1–98.2% of token throughput versus CC-off, with per-token latency (TPOT) within roughly 1.2–4.3% of baseline.

7.0

2026-09-22

CODING · 1 SOURCE · NVIDIA Developer

NVIDIA announces Topograph toolkit to publish cluster topology for topology-aware scheduling

NVIDIA Topograph is an open-source toolkit that discovers and normalizes cluster topology from cloud APIs and on‑prem fabrics, then publishes that model as Kubernetes node labels, Slurm configuration, or Slinky ConfigMaps so workload managers can perform topology-aware scheduling. It supports several cloud providers and on‑prem fabrics (InfiniBand, Spectrum‑X, Multi‑Node NVLink), integrates with DSX OS components such as DRA and KAI Scheduler, and is available from the dsx-ai-factory/topograph GitHub repo with Helm and native Slurm deployment options and simulation utilities.

6.0

2026-09-22

STARTUPS · 1 SOURCE · The Verge AI

Andreessen Horowitz to launch Horowitz Andreessen Academy with major tech partners

Venture firm Andreessen Horowitz (a16z) is creating the Horowitz Andreessen Academy, a one-year, highly selective program that will not grant degrees but offers short classes from tech leaders (including OpenAI CEO Sam Altman), co-op placements at partner companies, and more than $50,000 in compute credits. The academy launches with $42 million in funding led by a16z, 10 announced partners (Anduril, Anthropic, Coinbase, Google, Meta, NVIDIA, OpenAI, Palantir, Replit, Stripe), will admit about 50 students for a free inaugural fall 2027 cohort, and requires participants to relocate to San Francisco.

6.0

2026-09-22

MODELS · 1 SOURCE · NVIDIA Developer

NVIDIA launches DLSS 5 with 3D-Guided Neural Rendering; ACE and NVIGI gain new models

NVIDIA’s DLSS 5 adds 3D-Guided Neural Rendering that uses the game engine’s rendered frame as a strict base to add lifelike lighting and material detail while preserving artist-authored geometry and temporal stability; the compact model runs locally on GeForce RTX 50 Series GPUs at up to 4K and is available now in NBA 2K27. NVIDIA also updated ACE (Nemotron Speech 3.5 streaming ASR and Qwen3 TTS) and the In-Game Inferencing SDK with RTX Spark, Gemma4 integration, a Stable Diffusion plugin and llama.cpp performance improvements, and released RTX Mega Geometry 2.0 for streaming high-density LOD clusters.

8.0

2026-09-22

CODING · 1 SOURCE · NVIDIA Developer

NVIDIA updates Isaac ROS 5.0 to use rosidl::Buffer CUDA backend for zero-copy GPU transport

NVIDIA contributed a CUDA buffer backend to ROS 2 Lyrical and updated all nodes in NVIDIA Isaac ROS 5.0 to use rosidl::Buffer, enabling zero-copy GPU-resident message transport when runtime conditions allow. The post demonstrates an AI coding agent (migrate-node-to-rosidl-buffer skill) migrating a TensorRT-based Depth Anything 3 node, and verifies the optimized path with NVIDIA Nsight Systems and runtime checks that report backend_type == "cuda".

6.0

2026-09-21

COMPANIES · 1 SOURCE · NVIDIA Developer

NVIDIA Dynamo-Triton 26.07 enables TensorRT multi-device inference across GPUs

NVIDIA has enabled TensorRT multi-device inference in Dynamo-Triton release 26.07 (TensorRT multi-device is supported starting in TensorRT 11.0), allowing a single KIND_MODEL instance to own multiple GPUs and present one gRPC endpoint while using NCCL-backed distributed collectives. In a Cosmos 3 Nano video-generation demo, distributing a 44,160-token transformer across up to eight GPUs reduced end-to-end generation latency from 156.6s to 34.2s and produced a reported transformer RPC speedup of 6.09x; users can download Dynamo-Triton 26.07 from NGC and follow the TensorRT multi-device guide to reproduce the setup.

7.5

2026-09-21

RESEARCH · 1 SOURCE · NVIDIA Developer

Agent evaluation shifts to executable end-to-end and step-level scoring; NVIDIA Nemotron 3.5 Lightning demonstrates the approach

Agent evaluation is moving from scoring individual function calls to running agents inside executable environments that track state across multi-step tool use, combining step-level (process) scoring to locate failures with end-to-end (outcome) scoring to verify final task completion; metrics roll up through a fixed Benchmark→Trial→Task→Turn→Step hierarchy and favor executable checks over reference- or LLM-judge methods. NVIDIA presents this framework and reports Nemotron 3.5 Lightning achieving 86% accuracy on PinchBench while completing tasks ~30% faster than comparable models, and recommends domain-specific, state-gated production evaluations; reproducibility docs and a demo on build.nvidia.com are provided.

7.0

2026-09-21

MODELS · 1 SOURCE · NVIDIA Developer

NVIDIA Earth‑2 adds AI data‑assimilation tools (SDA, HealDA) for weather forecasting

NVIDIA says Earth‑2 now includes AI data‑assimilation tools that ingest proprietary and third‑party observations to update forecasts more frequently. Features include Score‑Based Data Assimilation (SDA) to constrain diffusion models such as CorrDiff and StormCast (NVIDIA reports a 54% wind‑speed RMSE reduction in a CorrDiff‑COSMO downscaling example) and HealDA, which maps heterogeneous observations onto a 1° HEALPix grid for rapid global state estimation.

7.0

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

2026-09-18

COMPANIES · 1 SOURCE · NVIDIA Developer

NVIDIA releases AIPerf, a multiprocess LLM benchmarking tool

NVIDIA AIPerf replaces GenAI-Perf with a ground-up multiprocess architecture designed to avoid client-side bottlenecks for high-concurrency LLM inference benchmarking. It supports 15+ endpoint types, public datasets and trace replay formats (ShareGPT, Mooncake, Baseten, WEKA/AgentX), configurable arrival patterns, and reports TTFT, ITL, latency percentiles and GPU telemetry when DCGM or pynvml are available.

6.0

2026-09-18

RESEARCH · 2 SOURCES · The Decoder · TechCrunch AI

Google DeepMind launches DeepMind Institute to study AGI safety, governance and risks

Google DeepMind has founded the DeepMind Institute (DMI), an interdisciplinary platform that brings together researchers from Google DeepMind, Google, and the global scientific community to research and debate artificial general intelligence (AGI) safety, governance, and risks such as cyberattacks or loss of control. Directors named in the announcement include Shane Legg, James Manyika and Demis Hassabis; DeepMind frames AGI as a system with the cognitive abilities of the human brain and discusses differing views on timing and definitions.

7.0

2026-09-17

REGULATION · 1 SOURCE · WIRED AI

WIRED podcast episode outlines plausible 'AI apocalypse' scenarios and safety debate

WIRED's Uncanny Valley podcast walks through three real-world scenarios experts worry about—hacked water supplies, bioweapons, and autonomous robots—and summarizes reactions from AI leaders (including Sam Altman and Dario Amodei) following a Salesforce event. The episode also highlights political attention to AI risks and recent industry developments such as an Anthropic researcher’s resignation.

5.0

2026-09-17

COMPANIES · 1 SOURCE · TechCrunch AI

Pinterest debuts Restyle beta — an AI tool to visualize home redecorating in your own photos

Pinterest is launching Restyle in beta in the U.S. and Canada, a consumer feature powered by Pinterest Intelligence that lets users upload a photo of a room and use AI to add, swap, erase, or restyle furniture, decor, lighting and finishes. Pinterest says Pinterest Intelligence runs on NVIDIA Blackwell GPUs and Dynamo together with open-source models and Pinterest-built tech; Restyle was teased at the Pinterest Presents event and will roll out more broadly next month.

6.0