{
  "id": 16953406,
  "title": "I tested Nvidia’s Nemotron 3.5 Bolt, and it changed what I thought my 16GB GPU could do",
  "url": "https://urgent.news/2026/10/28/i-tested-nvidias-nemotron-3-5-bolt-and-it-changed-what-i-thought-my",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-10-28T20:30:20.000Z",
  "source": {
    "name": "XDA Developers",
    "slug": "xda-developers",
    "url": "https://www.xda-developers.com/i-tested-nvidias-nemotron-35-bolt-and-it-changed-what-i-thought-my-16gb-gpu-could-do/"
  },
  "original_language": "en",
  "account": "Nvidia's Nemotron 3.5 Bolt AI model has proven to be more capable than initially expected on a 16GB RTX 4070 Ti Super GPU. The 74B model, with 9B active parameters and native 2-bit weights, can run entirely on the 24GB VRAM limit of the GPU. After some adjustments to the llama.cpp build, the model managed to process 19 tokens per second on average, with prompt processing at 110 tokens per second. The inclusion of LiLiCorr slowed things down, so speculative decoding was avoided.\n\nThe model's agentic capabilities were demonstrated through Hermes Agent reasoning traces, which were used to create a custom endpoint for llama.cpp. The agent was able to complete a five-step job, including finding installed software versions, checking GitHub for updates, finding a relevant pull request, writing a report, and saving the process as a skill. All of this was accomplished without human intervention, showcasing the potential of Nemotron 3.5 Bolt on a relatively modest GPU setup.",
  "summary": "A 16GB GPU can run agentic AI now, and it's surreal",
  "key_points": [
    "Nvidia's Nemotron 3.5 Bolt AI model outperformed expectations on 16GB RTX 4070 Ti Super GPU",
    "Hermes Agent demonstrated model's agentic capabilities by completing five-step job autonomously"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "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."
}