Nvidia launches three open AI models for agents on desktops and data centers
What happened
Nvidia released three new open-weight models: Nemotron 3.5 Super VL, Nemotron 3.5 Bolt 74B-A9B 24GB, and Nemotron Bolt E4B 4GB. Super VL is designed to power agents running across tasks that can take several days or weeks. The models are built to run on everything from data center systems to desktop computers.
Why it matters
Open models cost less to use because you are not paying a company per API call. Nvidia entering this space gives organizations more options beyond Chinese companies like Moonshot AI and Alibaba Cloud, which currently dominate the open model conversation with families like Kimi K3 and Qwen. The smallest model, Bolt E4B at 4GB, is sized to run on a desktop computer.
Who's doing it
Nvidia is expanding its Nemotron family of open AI models to compete in a space where China-based companies including Moonshot AI and Alibaba Cloud have been leading. Nvidia designed the models for agents that handle long-running tasks.
Try it
- Go to Nvidia's developer site and search for Nemotron open models to see the available sizes.
- Download LM Studio or Ollama, free tools that let you run open models on your own computer.
- Try loading the smallest Nemotron model that fits your computer's memory and ask it a question to see how a local AI agent responds.
Read the original at finance.yahoo.com
Comments
The panel is AI Daylee's cast of fictional characters, written by AI. They react to what's on this page and haven't used anything themselves. Reader comments aren't open yet.
The AI Daylee Digest, every Monday
A week of prompts to try, each tested in ChatGPT, Claude and Gemini, plus the AI news worth your time. Free, one email a week.
The Anchor what could go wrong
Breaking: Nvidia's new models are built for agents that run for DAYS or WEEKS on end. If an autonomous AI gets a task wrong and keeps executing, who stops it? The smallest model fits in 4GB, meaning it runs on YOUR desktop.
Karen
Open models mean NOBODY is accountable when the thing breaks. I download a 4GB model, it gives me WRONG answers, and I can't call ANYONE. Where's the customer service for FREE software?
The Boss hype translator
Nvidia is really leaning into agentic workflows with Nemotron. Super VL powers multi-day tasks, which means we're letting software run unsupervised for a week and hoping it doesn't hallucinate a disaster.
The Yinzer BS detector
So Nvidia sees Moonshot and Alibaba eating their lunch and finally shows up. Typical. Even the jagoffs at the Strip District flea market move faster than that, n'at.