Runway releases Praxis‑1 world action model with open weights
Runway has made Praxis‑1, a 27-billion-parameter model designed to predict robot video and actions jointly, generally available with Apache 2.0 open weights.
AT A GLANCE
Key points
- 27B mixture-of-experts diffusion transformer
- Apache 2.0 weights and code
- GitHub, Hugging Face and hosted Reactor access
What is changing?
Runway made Praxis‑1 generally available on November 11. The 27-billion-parameter mixture-of-experts diffusion transformer predicts video frames and robot actions within one model. The weights plus inference and fine-tuning code are available under Apache 2.0 on GitHub and Hugging Face, alongside LeRobot and vLLM‑Omni integrations and hosted access through Reactor.
In Runway’s reported RoboLab‑120 results, Praxis‑1 reached a 46.2% task success rate across 1,200 episodes. Runway also says fine-tuning on Google RT‑1 data increased SimplerEnv success from 52.4% to 80.0%. These figures are provider-reported. The release targets hardware including H100/B200, RTX 5090/PRO 6000 and Jetson Thor.
Explore the original sources
Source date: 11 November 2026.
Published by AI Radar: 12 Nov 2026, 01:07 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…