Skip to content

Dataset · agibot-world

Real-world data for AgiBot World 2026 robots

agibot-world/AgiBotWorld2026

Corpus of real scenes with images and video for training and evaluating embodied AI and robotics.

What's inside

A dataset for embodied intelligence research: real scenes, images and video. According to the card, it covers key robotics directions and is intended for the next generation of models. Metadata language is English, volume is from 100k to 1M units.

What you can do with it

  • Training VLA models and robotic policies
  • Testing embodied agents on real scenes
  • Pretraining video and image models for robotics
Tasks
Robotics
Languages
English
Size
100K–1M rows
Downloads this month
2.1M
Likes
115

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("agibot-world/AgiBotWorld2026")