MindPattern
← The Wire
Source trail

NVIDIA

Public MindPattern findings, entities, and graph evidence that cite this source.

Findings
40
All-time hits
1
High value
1
Last seen
2026-02-22

Connected entities

NVIDIA AI BlogNVIDIA Confidential Computing Now Powers Apple's Private Cloud Compute Beyond Apple's Own Data CenteNVIDIA Technical Blog — Practical Security Guidance for Sandboxing Agentic WorkflowsEnforce NVIDIA's three non-negotiable agent sandbox controls: egress allowlist, workspace write limiNVIDIA NewsroomNVIDIA Formally Launches Vera Rubin Data Center Platform at GTC Taipei — Six Co-Designed Chips for NJensen Huang: Vera Rubin Platform Now in Full Production — 10x Inference Cost Reduction, 2x Supply CJensen Huang + Satya Nadella on Stage Together: 'Unmetered Intelligence to Every Home and Every DeskNVIDIA Open-Sources Major Collection of Agent Tools for Physical AI — Nemotron Agentic Safety DataseNVIDIA Vera Rubin Platform Now in Production — 10x Agentic Throughput vs Blackwell, Shipping Fall 20Jensen Huang: NVIDIA Vera CPU Unveiled — First Processor Purpose-Built for AI Agents, Anthropic and NVIDIA Cosmos 3: First Fully Open Physical AI Omnimodel — #1 on 7+ Robotics BenchmarksNVIDIA Technical BlogNVIDIA Launches Verified Agent Skills Framework: SkillSpector Security Scanning and Cryptographic SiNVIDIA BlogJensen Huang at Dell World: 'Demand Is Going Parabolic' — Agentic AI Driving 100x-1000x Compute Grow

Related findings

  1. 2026-07-01 / SOURCESNVIDIA Expands Nemotron 3 Open Models with Speech, Multimodal RAG, and Safety — Plus Palantir Sovereign DeploymentOn June 30 NVIDIA broadened its open-weight Nemotron 3 family with new models for speech, multimodal retrieval-augmented generation, and safety, alongside open datasets and libraries aimed at agentic AI. A day earlier Palantir launched an engine to run Nemotron open models in sovereign/air-gapped US government environments, and Nemotron 3 Ultra already has day-0 vLLM support — a concrete open-weights path for builders needing on-prem or regulated deployment.
  2. 2026-06-30 / NEWSFirefly Aerospace Runs an NVIDIA Jetson in Lunar Orbit for the First TimeFirefly Aerospace operated an NVIDIA Jetson edge module in lunar orbit, marking what NVIDIA says is the first time its edge AI hardware has run that far from Earth. It demonstrates on-device inference for autonomous spacecraft operations where round-trip latency to Earth is prohibitive. A notable proof point for ruggedized edge AI beyond terrestrial deployments.
  3. 2026-06-30 / SOURCESNVIDIA's June 2026 DGX Spark Update Makes Fully-Local Agents PracticalNVIDIA's June 2026 DGX Spark software release ships automated four-node clustering via a new Cluster Assistant (enabling ~700B models locally), a 2.6x throughput gain on Qwen3.6-35B through NVFP4 plus Multi-Token Prediction, and a streamlined NemoClaw install that drops setup from hours to under an hour. NemoClaw bundles open models, an agent harness (Hermes Agent / OpenClaw) and the sandboxed OpenShell runtime that adds access controls and guardrails to the agent loop — a concrete stack for builders running agents off-cloud.
  4. 2026-06-25 / DISPATCHNVIDIA and AWS Make GPU Vector Indexing the Default in OpenSearch Serverless via cuVS — Up to 10x Faster at a Quarter of the CostNVIDIA and AWS announced (June 23) that the NVIDIA cuVS library now powers GPU-accelerated vector indexing as the default in Amazon OpenSearch Serverless, claiming up to 10x faster index builds at roughly a quarter the cost versus CPU-only, making billion-scale vector databases buildable in under an hour. The same announcement debuts EC2 G7 instances on NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, part of a plan to deploy over 1M GPUs in 2026. For RAG and agent builders, this directly attacks the cost and latency of the retrieval layer at production scale.
  5. 2026-06-25 / AGENTSNVIDIA brings 24/7 autonomous AI agents to telecom operationsAt DTW Ignite 2026, NVIDIA detailed 'trusted' always-on AI agents for telecom network operations, positioning continuous autonomous agents to monitor and act on carrier infrastructure. It is another vertical-specific agent deployment (joining healthcare, retail, and advertising rollouts) where the selling point is reliability and trust for round-the-clock execution rather than raw capability. A primary-source vendor announcement signaling telecom as a maturing agent vertical.
  6. 2026-06-23 / SOURCESNVIDIA Drops a Massive Open Bundle: Nemotron Models, 10T Tokens, and an LLM Router BlueprintOn June 22 NVIDIA released open models, datasets, and tools across the Nemotron family — including training data and code for the Llama Embed Nemotron 8B embedding model (on the MMTEB leaderboard), an updated LLM Router blueprint that auto-directs requests to the best model for a job, and the dataset behind Nemotron Speech ASR. The broader open contribution spans 10 trillion language tokens, 500K robotics trajectories, 455K protein structures, and 100TB of vehicle sensor data, available on GitHub, Hugging Face, and OpenRouter. Verified on the NVIDIA blog.
  7. 2026-06-20 / PROJECTSNVIDIA Releases a Major Open-Source Collection of Agent Tools and Skills for Physical AINVIDIA announced a large open-source release of agent tools and skills targeted at Physical AI — robotics and embodied agents — via its newsroom. The bundle gives developers reusable building blocks for agents that act in the physical world rather than just text. It signals NVIDIA pushing the agent-tooling layer down toward robotics.
  8. 2026-06-18 / SOURCESNVIDIA Open-Sources Cosmos World Models and 1,700+ Hours of Physical-AI Data on Hugging FaceOn June 16 NVIDIA released a major open bundle on GitHub and Hugging Face, including Cosmos open world foundation models (Cosmos Reason 2, a leaderboard-topping reasoning VLM), over 1,700 hours of multi-geography driving data, and a 455K synthetic protein-structure dataset. The release targets robotics, autonomous vehicles, and physical-AI research with permissively available weights and data rather than API-only access. For builders outside frontier-chat, it's a rare large open drop aimed squarely at embodied and simulation work.
  9. 2026-06-17 / PROJECTSNVIDIA Ships Cosmos3-Super, a 64B World Model Coupling Action Trajectories With Video+AudioNVIDIA released Cosmos3-Super, a 64B physical-AI world model split into a 32B reasoner plus a 32B generator that couples action trajectories with synchronized video and audio generation, aimed at robotic policy development and synthetic data. It natively understands and produces text, images, video, ambient sound, and physical actions, ranking first among open-weight options on several physical benchmarks. For builders in robotics and simulation, it's a notable open foundation for sim-to-real data generation rather than another chat model.
  10. 2026-06-12 / DISPATCHNVIDIA Argues Robotaxi Safety Must Be 'Built In, Not Bolted On' with HALOS OSNVIDIA published a piece on its HALOS OS framework arguing that robotaxi safety has to be architected into the autonomy stack from the start rather than layered on afterward, as driverless services now operate in dozens of cities. It's a positioning play for NVIDIA's safety platform amid scaling robotaxi deployments. Signals where the autonomy-stack vendor competition is heading.
  11. 2026-06-10 / NEWSNVIDIA Confidential Computing Now Powers Apple's Private Cloud Compute Beyond Apple's Own Data CentersNVIDIA announced that its GPUs with Confidential Computing are now used for confidential inference inside Apple's Private Cloud Compute (PCC), as Apple expands PCC beyond its own data centers onto Google Cloud. Unveiled alongside Apple's WWDC announcements, this marks a notable hardware-security milestone: frontier AI inference running with hardware-enforced confidentiality on third-party infrastructure. It signals a path for privacy-preserving cloud AI that other vendors will be pressured to match.
  12. 2026-06-08 / DISPATCHNVIDIA and LG Group Build a Joint 'AI Factory' for Physical AI and MobilityNVIDIA and LG Group announced an AI factory to accelerate LG's robotics, autonomous driving, data-center, and GPU-cloud businesses. The deal extends NVIDIA's 'AI factory' framing — full-stack compute buildouts dedicated to a single enterprise's AI workloads — into physical AI and mobility. It's another data point in the global capex race to stand up dedicated AI infrastructure outside the hyperscalers.
  13. 2026-06-07 / PROJECTSNous Research's Hermes Agent Surpasses 140K Stars With Self-Improving, Skill-Creating CLI LoopHermes Agent from Nous Research has crossed 140,000 GitHub stars in under three months (some trackers now cite ~176k), positioned as a self-improving CLI agent with persistent memory, automated skill creation, sandboxed code execution, and support for 300+ models. NVIDIA spotlighted it running locally on RTX PCs and DGX Spark, underscoring an on-device self-improvement narrative. For builders, it's one of the most concrete open implementations of an agent that writes its own skills.
  14. 2026-06-06 / AGENTSNVIDIA Publishes Cosmos 3 World Models and Modular Agent Skills for Physical AIOn June 4, 2026, NVIDIA released physical-AI research tooling built on Cosmos 3 world models plus modular 'agent skills' that integrate with Omniverse and Isaac Sim. The release brings the agent-skills abstraction into robotics/simulation workflows, letting developers compose reusable, swappable capabilities for embodied agents. It extends the skill-marketplace pattern from software agents into the physical-AI and humanoid-robotics domain.
  15. 2026-06-06 / SKILLSEnforce NVIDIA's three non-negotiable agent sandbox controls: egress allowlist, workspace write limits, config-file lockdownNVIDIA's 2026 sandboxing guidance defines three mandatory controls for autonomous coding agents: (1) network egress allowlists via HTTP proxy/IP/port so the agent calls only permitted APIs and all other outbound traffic alerts; (2) workspace write restrictions that exclude dotfiles and auto-executing config directories; (3) configuration-file protection that blocks all agent modification of hooks, MCP server defs, and IDE extensions regardless of approval level — manual user edits only. Reference implementation is NVIDIA OpenShell, governed by declarative YAML policies.
  16. 2026-06-02 / SKILLSNVIDIA Formally Launches Vera Rubin Data Center Platform at GTC Taipei — Six Co-Designed Chips for Next-Gen AIJensen Huang formally launched the NVIDIA Vera Rubin data center platform at GTC Taipei (co-located with Computex, June 2-5), comprising six co-designed chips: Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch. The platform is the successor to Blackwell/Grace Hopper, designed to slash both AI training time and inference token costs through extreme cross-chip co-design. This positions NVIDIA's next-generation stack for large-scale agentic and frontier model workloads.
  17. 2026-06-02 / VOICESJensen Huang: Vera Rubin Platform Now in Full Production — 10x Inference Cost Reduction, 2x Supply Chain Scale vs BlackwellAt Computex, Huang confirmed the Vera Rubin platform has entered full production with a supply chain twice as large as Grace Blackwell. The five purpose-built racks operate as one AI supercomputer delivering up to 10x reduction in inference token cost and 4x reduction in GPUs needed to train MoE models vs. Blackwell. AWS, Google Cloud, Microsoft, and OCI will deploy Vera Rubin instances in H2 2026. For builders: if inference costs drop 10x, the economics of running persistent AI agents changes fundamentally.
  18. 2026-06-02 / VOICESJim Fan: GR00T N2 Vision-Language-Action Model Ranks #1 on Both MolmoSpaces and RoboArena BenchmarksNVIDIA's next-generation GR00T N2 VLA model — co-led by Jim Fan as head of NVIDIA's embodied AI research — achieved the #1 ranking on both the MolmoSpaces and RoboArena robot benchmarks with 2x the success rate of leading competitors. Announced alongside the Isaac GR00T reference robot at Computex, this represents NVIDIA's strongest claim yet in the physical AI software stack. For builders in robotics, GR00T N2 is now the model to beat for humanoid manipulation tasks.
  19. 2026-06-02 / VOICESJensen Huang + Satya Nadella on Stage Together: 'Unmetered Intelligence to Every Home and Every Desk with Windows'Huang stood next to Nadella at Computex and declared 'the PC is being reinvented.' Nadella responded: 'Our goal is to deliver unmetered intelligence to every home and every desk with Windows,' calling RTX Spark 'a real breakthrough towards that vision.' This is the first time both CEOs have jointly announced a consumer computing product — signaling a deep NVIDIA-Microsoft alliance for agentic AI on consumer hardware that could reshape the Windows ecosystem.
  20. 2026-06-01 / SKILLSNVIDIA Open-Sources Major Collection of Agent Tools for Physical AI — Nemotron Agentic Safety Dataset, NeMo RL, Alpamayo 2 Super 32BReleased May 31 alongside Computex announcements, NVIDIA published a major collection of open-source agent tools for building, testing, and running physical AI systems. Includes the Nemotron Agentic Safety Dataset, NeMo RL reinforcement learning libraries, NeMo Evaluator, and the Alpamayo 2 Super 32B vision-language-action model for autonomous vehicles.
  21. 2026-06-01 / DISPATCHNVIDIA Vera Rubin Platform Now in Production — 10x Agentic Throughput vs Blackwell, Shipping Fall 2026NVIDIA announced at Computex that Vera Rubin is in full production on TSMC 3nm, combining an 88-core Vera CPU with next-gen Rubin GPU. The platform delivers 10x agentic AI throughput, 10x reduction in inference token cost, and 4x reduction in GPUs needed to train MoE models versus Blackwell. AWS, Google Cloud, Microsoft, OCI, CoreWeave, Lambda, and Nebius will be among the first to deploy Vera Rubin instances in late 2026.
  22. 2026-06-01 / REDDITNVIDIA Alpamayo: New Open Model Family for Autonomous Vehicle Development Announced at ComputexNVIDIA unveiled the Alpamayo model family alongside Nemotron 3 Ultra and Cosmos 3, specifically targeting autonomous vehicle perception and planning. Part of the broader open-source push that includes 100TB of vehicle sensor data. Alpamayo joins Isaac GR00T (robotics) and Clara (biomedical) as domain-specific model families, signaling NVIDIA's strategy of verticalizing open foundation models by industry rather than releasing one-size-fits-all architectures.
  23. 2026-06-01 / REDDITNVIDIA Open-Sources Massive Multimodal Training Data: 10 Trillion Language Tokens, 500K Robot Trajectories, 455K Protein StructuresAlongside the Computex model announcements, NVIDIA released one of the world's largest open multimodal data collections: 10 trillion language training tokens, 500,000 robotics trajectories, 455,000 protein structures, and 100 terabytes of vehicle sensor data. The dataset spans the Nemotron, Cosmos, Isaac GR00T, Clara, and Alpamayo model families. For builders, the robotics trajectories and protein structures represent previously gated data categories now accessible under open licenses.
  24. 2026-06-01 / REDDITNVIDIA Announces Nemotron 4 Development via 8-Lab Coalition — First Open Frontier Model Co-Developed on DGX CloudAt Computex, NVIDIA revealed the Nemotron Coalition's first joint model will underpin the Nemotron 4 family. The coalition (Mistral AI, Perplexity, Cursor, LangChain, Black Forest Labs, Reflection AI, Sarvam, and Thinking Machines Lab) co-develops on DGX Cloud infrastructure. NVIDIA's SEC filing disclosed a $26 billion five-year commitment to open-weight AI — the largest financial commitment to open-source AI ever made. Cursor's inclusion signals the coding agent market is now a first-class frontier model consumer.
  25. 2026-06-01 / VOICESJensen Huang: 'AI Is Now a GDP Generator' — Claims Agentic AI Triples Developer Productivity to $9T OutputAt GTC Taipei, Jensen Huang declared 'AI is now a profit generator. AI is now a GDP generator,' claiming agentic AI enables software developers to triple productivity — generating $9 trillion in output from $3 trillion in salaries. He unveiled Vera Rubin, a multi-rack pod-scale system integrating GPUs, CPUs, storage, and networking for agentic inference with NVLink 72 and advanced liquid cooling. NVIDIA's 150 Taiwan supply chain partners enabled rapid production. This frames agentic AI as an economic multiplier, not just a cost cutter.
  26. 2026-06-01 / VOICESJensen Huang: NVIDIA Vera CPU Unveiled — First Processor Purpose-Built for AI Agents, Anthropic and OpenAI Are Launch CustomersJensen Huang announced Vera, a standalone data center CPU with 88 custom Olympus cores, 227 billion transistors, and 1.2TB/s LPDDR5X memory bandwidth, designed specifically for agentic AI workloads. NVIDIA claims 1.8x faster task completion than x86 CPUs for agent inference. Initial customers include Anthropic, OpenAI, and SpaceX, with mass shipments beginning fall 2026. This is the first CPU designed around agent workloads rather than general compute, marking a new hardware category.
  27. 2026-06-01 / SOURCESNVIDIA Launches Cosmos Coalition With Runway, Black Forest Labs, Skild AI for Open World ModelsAlongside Cosmos 3, NVIDIA launched the Cosmos Coalition — a consortium of AI labs and robotics companies including Agile Robots, Black Forest Labs, Generalist, LTX, Runway, and Skild AI committed to advancing open world models for physical AI. This is NVIDIA's play to make Cosmos the default foundation for the emerging robotics/simulation ecosystem, similar to how CUDA became the default for GPU compute.
  28. 2026-06-01 / SOURCESNVIDIA Cosmos 3: First Fully Open Physical AI Omnimodel — #1 on 7+ Robotics BenchmarksCosmos 3 is a single architecture trained on 20 trillion multimodal tokens that natively reasons over images, generates simulated worlds, and produces robot action sequences (joint angles, trajectories). Available in Super (32B) and Nano (8B) variants under open weights. NVIDIA also launched the Cosmos Coalition with Black Forest Labs, Runway, and Skild AI to advance open world models. For builders in robotics or simulation: this collapses the stack from separate vision, planning, and action models into one downloadable omnimodel.
  29. 2026-05-27 / AGENTSNVIDIA Hermes Agent Hits 140K GitHub Stars: Self-Evolving Skills on DGX Spark with Qwen 3.6 at 1 PFLOPNVIDIA published a developer blog on May 13 detailing the Hermes Agent framework by Nous Research, which crossed 140,000 GitHub stars in under three months and ranks as the most-used agent globally on OpenRouter. Hermes writes and refines its own skills from experience, persists memory across sessions, delegates to short-lived sub-agents for parallel sub-tasks, and runs scheduled workflows via built-in cron. On NVIDIA DGX Spark (128GB unified memory, 1 PFLOP AI performance), Hermes runs Qwen 3.6 35B on ~20GB while surpassing 120B-parameter models — the 27B variant matches accuracy of 400B models at 1/16th the size.
  30. 2026-05-23 / AGENTSNVIDIA Launches Verified Agent Skills Framework: SkillSpector Security Scanning and Cryptographic Signing for Enterprise Agent DeploymentNVIDIA published the Verified Agent Skills framework on May 22, 2026, introducing a publication pipeline that scans, signs, and documents portable agent skill packages. SkillSpector checks both conventional software risks (vulnerable dependencies, credential access, data exfiltration) and agent-specific risks (hidden instructions, prompt injection, trigger abuse, tool poisoning). Each skill ships with a machine-readable skill card documenting provenance, licensing, dependencies, and known limitations, with cryptographic signatures covering every file in the skill directory.
  31. 2026-05-23 / SKILLSHermes Agent v0.10 Ships Closed Learning Loop: Agent Creates Reusable Markdown Skill Files from Experience, Benchmarked at 40% Task Time ReductionNous Research's Hermes Agent v0.10 (MIT license) ships 118 bundled skills, three-layer memory, and six messaging integrations. The core differentiator is the closed learning loop: when the agent solves a task, it writes a reusable Markdown skill file, stores the outcome in persistent memory, and adjusts its approach next time. TokenMix benchmarks show self-created skills cut research task time by 40% versus a fresh instance. At ~$0.30 per complex task on budget models and $5-10/month for always-on VPS hosting, the economics are viable for solo builders running personal agent infrastructure.
  32. 2026-05-21 / RESEARCHNVIDIA Nemotron 3 Nano Omni: Open 30B MoE Model With 3B Active for Edge AI AgentsNVIDIA released Nemotron 3 Nano Omni, an open multimodal model with 30B total parameters but only 3B active per forward pass via mixture-of-experts, designed for edge deployment in agentic workflows. The model tops six benchmarks across document intelligence, video understanding, and audio comprehension, delivering 9x higher throughput than comparable open multimodal models with accuracy jumping from 11.1 to 47.4 on the OSWorld GUI agent benchmark.
  33. 2026-05-20 / AGENTSJensen Huang at Dell World: 'Demand Is Going Parabolic' — Agentic AI Driving 100x-1000x Compute GrowthNVIDIA CEO Jensen Huang appeared at Dell Technologies World keynote on May 18, declaring 'now we have, for the very first time, useful AI' as agentic systems move from generative to reasoning and planning. He stated compute demand has grown 100x to 1000x depending on workload. Dell and NVIDIA co-announced the Dell AI Data Platform as long-term memory infrastructure for agents, with Dell claiming enterprises are seeing 10-30x improvements (not incremental 10-30%) when reimagining workflows with agentic AI.
  34. 2026-05-19 / NEWSNVIDIA Hand-Delivers First Vera CPUs to Anthropic, OpenAI, SpaceXAI, and Oracle — 88 Custom Cores for Agentic AINVIDIA VP Ian Buck hand-delivered the first Vera CPUs to Anthropic, OpenAI, SpaceXAI, and Oracle Cloud Infrastructure over the weekend. Vera is NVIDIA's first custom CPU, featuring 88 Olympus cores, 1.2 TB/s memory bandwidth, and 50% faster per-core performance than predecessors — purpose-built for agentic AI workloads. Oracle plans to deploy hundreds of thousands of Vera CPUs, and Jensen Huang positioned the standalone CPU as NVIDIA's next multi-billion-dollar business line.
  35. 2026-05-19 / VOICESJensen Huang CMU Commencement: 'Largest Technology Infrastructure Buildout in Human History'Jensen Huang delivered the commencement keynote at Carnegie Mellon University on May 10, telling the Class of 2026: 'I cannot imagine a more exciting time to begin your life's work.' He described AI as driving 'the largest technology infrastructure buildout in human history' and a 'once-in-a-generation opportunity to reindustrialize America and restore the nation's capacity to build.' The contrast with Eric Schmidt's booed AI commencement at Arizona eight days later is stark — Huang's CMU audience was receptive.
  36. 2026-05-15 / SKILLSNVIDIA Partners With Nous Research on Hermes: Qwen 3.6-27B Outperforms Previous 397B Models on Agentic Coding — Optimized for Always-On Local Inference on DGX Spark's 128GB Unified MemoryNVIDIA's RTX AI Garage announced Hermes Agent optimization for DGX Spark (128GB unified memory, 1 petaflop AI) and RTX PCs on May 13. The key technical claim: Qwen 3.6 27B and 35B parameter models now outperform their previous-generation 120B and 400B counterparts on agentic coding benchmarks, making high-quality local agents practical on consumer and prosumer hardware. Hermes is provider- and model-agnostic, designed for always-on 24/7 local use. Combined with Hermes's self-evolving skill architecture, this positions local agents as a serious alternative to cloud APIs for developers who want persistent, private, and rate-limit-free agent access.
  37. 2026-05-15 / VOICESJensen Huang CMU Commencement: 'People Using AI Will Replace People Without AI Skills' — Calls It Largest Infrastructure Buildout in HistoryNVIDIA CEO Jensen Huang delivered the keynote at Carnegie Mellon's 128th commencement on May 10, telling graduates they are 'entering the world at an extraordinary moment.' His key prediction: AI will not replace individual workers, but people who use AI well will replace people without AI skills. Huang described AI as driving 'the largest technology infrastructure buildout in human history' and a 'once-in-a-generation opportunity to reindustrialize America.' He received an honorary Doctor of Science and Technology from CMU. For builders: the NVIDIA CEO is framing AI skills as career insurance, not career threat.
  38. 2026-05-13 / DISPATCHNVIDIA Partners with AlphaGo Architect David Silver's Ineffable Intelligence on RL InfrastructureNVIDIA announced a collaboration with Ineffable Intelligence, the London AI lab founded by DeepMind's former RL lead David Silver, to build reinforcement learning infrastructure at scale using Grace Blackwell and Vera Rubin chips. Ineffable raised a record $1.1B seed co-led by Sequoia and Lightspeed in April. Silver states the harder AI problem is systems that 'discover new knowledge for themselves.'
  39. 2026-05-13 / SOURCESNVIDIA Releases Clara AI Open Models for Drug Discovery — La-Proteina, ReaSyn v2, KERMT, RNAProNVIDIA released new open Clara AI models on May 13 targeting healthcare and life sciences: La-Proteina for designing large atom-level-precise proteins, ReaSyn v2 for ensuring AI-designed drugs are practical to synthesize, KERMT for high-accuracy computational safety testing, and RNAPro for predicting complex 3D RNA molecular shapes. These join the broader Nemotron 3 family push to provide open models across every industry vertical. For builders working in biotech or scientific computing, these represent ready-to-use checkpoints for specialized domains.
  40. 2026-05-13 / PROJECTSNVIDIA Unveils Ising Quantum AI Models, Nemotron Speech, and Nemotron RAG in May 13 Open-Source ReleaseNVIDIA released a suite of open models on May 13 including Ising — the world's first open-source quantum AI model family for accelerating quantum processor development — plus Nemotron Speech (low-latency real-time ASR) and Nemotron RAG (multimodal embedding and reranking VLMs for document retrieval). All models, training data, and reference implementations are available through GitHub, Hugging Face, and NVIDIA's developer platforms.
Open latest cited source