---
format: "aidr-story-markdown/v1"
id: "e5a3d93ab4ee3102f6171584c8f74728ad58b9f364a79a5c233fb9aded3e08e4"
canonical_url: "https://aidr.today/e5a3d93a?lang=en"
title: "Nvidia's model GR00T 2 watches humans on video, then imagines itself as the robot"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-10T18:41:52.000Z"
category: "Models"
topics: ["nvidia","agent"]
source_urls: ["https://the-decoder.com/nvidias-model-gr00t-2-watches-humans-on-video-then-imagines-itself-as-the-robot/"]
summary: "Nvidia's new AI model GR00T 2 transfers human hand movements from videos to robotic arms. The system was trained on about 400,000 hours of video footage. Success rates range from 10 to 80 percent depending on the task."
---

# Nvidia's model GR00T 2 watches humans on video, then imagines itself as the robot

> [Open the canonical story](<https://aidr.today/e5a3d93a?lang=en>)

**Published:** 2026-11-10T18:41:52.000Z
**Category:** Models
**Topics:** nvidia, agent

## Summary

Nvidia's new AI model GR00T 2 transfers human hand movements from videos to robotic arms\. The system was trained on about 400,000 hours of video footage\. Success rates range from 10 to 80 percent depending on the task\.

## Sources

- [Story source](<https://the-decoder.com/nvidias-model-gr00t-2-watches-humans-on-video-then-imagines-itself-as-the-robot/>)