Dyna Robotics unveils DYNA-2 World-Action Model, demonstrating first true scaling law in robotics powered entirely by human data

Dyna Robotics introduced DYNA-2, a new robot foundation model trained on over a million hours of human-recorded video. This approach uses pure video data, unlike previous models relying on limited action datasets, setting a new benchmark for physical AI development.
Key takeaways
- New robot model trained on extensive human video data
- Breaks from previous reliance on limited action datasets
- Establishes a novel scaling method for physical AI
- Potential for more capable and adaptable robotic systems
Why it matters
This advancement could lead to more adaptable and intuitive robotic assistants capable of understanding and performing complex tasks. For professionals, it signals a future where AI tools in physical environments are more capable of learning from real-world human interactions.
Try this on SynaBot
Related AI assistants, prompts, and tools from the SynaBot catalog.
- Role Model AIRole Model AI — Virtual assistant with 3D avatars, phone integration, and Fortnite connectivity. It sits in the 3d category and is built to generate or edit 3D assets, prototypes, and visualizations for product, game, or architectural workflows.
- Mental Models AIMental Models AI offers AI-driven coaching and bias recognition to help data and analytics professionals make smarter business decisions, generate insights, and optimize reporting workflows.
- Looker ModelerLooker Modeler provides a robust semantic layer to define metrics and dimensions centrally. Ensure consistent data interpretation across all reports and dashboards. Empower business users with trusted, self-service analytics.


