{
  "id": 14725967,
  "title": "Nvidia’s new Nemotron 3.5 Super VL is another step forward for US open-weight AI",
  "url": "https://urgent.news/2026/10/15/nvidias-new-nemotron-3-5-super-vl-is-another-step-forward-for-us-open",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-15T15:00:00.000Z",
  "source": {
    "name": "ZDNet",
    "slug": "zdnet",
    "url": "https://www.zdnet.com/innovation/nvidia-nemotron-3-5-super-vl-model-us-open-weight-ai/"
  },
  "original_language": "en",
  "account": "Nvidia's latest open-weight AI model, Nemotron 3.5 Super VL, signifies a significant step in the ongoing competition for top-tier US-made open models. This new 120-billion-parameter model targets the growing demand for agentic work by delivering unmatched speed and precision for specialized, always-on agents. According to Nvidia, it is a staggering 5x faster than its closest competitor, setting it apart from other open models like InclusionAI's Ling 3.0 Flash and DeepSeek's V4.1 Flash, which placed second and third in benchmark tests. The model demonstrates exceptional accuracy for specialized agents, with a notable 44% increase in medical insurance claims, outperforming Claude Opus 5. What sets open-weight models apart from their closed counterparts is their adaptability. Businesses can post-train open models on their proprietary data, allowing for personalized and secure applications. To further enhance its offerings, Nvidia has introduced two new edge models: Nemotron 3.5 Bolt 74B-A9B 24GB and Nemotron 3.5 Bolt E4B 4GB. Edge AI refers to models designed to operate independently, without relying on external infrastructure. While it remains to be seen whether Nvidia's new model will surpass open-weight or proprietary models in practice, the benchmark results indicate efficiency gains for now. This groundbreaking release is part of a broader shift in the AI landscape, as open models from Chinese companies like DeepSeek and Moonshot.ai have traditionally represented the promise and risk of cost-effective, customizable models compared to their closed, pricier alternatives. However, recent advancements from French lab Mistral and Reflection AI, which Nvidia is in talks to acquire, indicate a closing gap in the open-weight wars. The industry is increasingly embracing open-weight models, following the OpenAI Hugging Face breach, which eliminated the previous belief that only open models posed security risks. If US-made open models can outperform Chinese counterparts, it could help propel open models into new markets where they have yet to establish a strong foothold, providing a viable alternative to foreign technologies in terms of cybersecurity.",
  "summary": "Nvidia’s new models join a wave of open-weight releases from US labs on the heels of the Hugging Face hack.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "SiliconANGLE",
        "title": "Nvidia expands Nemotron LLM family with hyper-efficient models for agentic systems",
        "url": "https://urgent.news/2026/10/15/nvidia-expands-nemotron-llm-family-with-hyper-efficient-models-for",
        "published": "2026-10-15T17:10:27.000Z"
      }
    ]
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}