Archive · 12 Jun 2025
Meta's V-JEPA 2 AI: Mimicking Human Understanding
Imagine teaching a robot to understand the world like you do!
Meta has created V-JEPA 2, a special AI that learns by watching videos.
It sees how things move and interact, like a ball falling to the ground.
This AI can predict what will happen next, even when it hasn't been told what to do.
It was tested on robots that learned to pick up objects and move them without any instructions.
This new AI could make robots much smarter and more helpful in the future!
Meta has unveiled V-JEPA 2, an advanced AI model designed to understand and predict the physical world like humans.
The model is trained on over a million hours of video data, allowing it to observe everyday actions and their outcomes.
V-JEPA 2 can learn by watching, without needing extra labels or instructions.
Meta integrated V-JEPA 2 into robots, which were able to perform tasks like picking up and moving objects.
The model is a step towards Advanced Machine Intelligence (AMI) and the development of more capable AI systems.
- Who
- Meta developed V-JEPA 2.
- What
- Meta launched V-JEPA 2, an advanced AI model designed to mimic human understanding of the physical world.
- When
- The launch of V-JEPA 2 was recently announced.
- Where
- The model was developed by Meta.
- Why
- V-JEPA 2 aims to improve AI's ability to understand and predict the physical world, advancing towards more capable AI systems.
- Model Name
- V-JEPA 2
- Developer
- Meta
- Training Data
- Over one million hours of video
- Parameters
- 1.2 billion
- Focus
- Understanding and predicting physical world
- Goal
- Advancing toward Advanced Machine Intelligence (AMI)