Runway open-sources Praxis-1, a 27B video-action model topping NVIDIA's robot leaderboard
runwayml · x · 2026-11-11
Runway Releases Open-Weight World Action Model Praxis-1
Runway announced Praxis-1, a 27B-parameter mixture-of-experts diffusion transformer that predicts video and actions together. The core idea: transfer Runway's large-scale video pretraining into real robot control, sidestepping the scarcity of real-world robot data.
Results & capabilities:
- Ranks #1 on NVIDIA's RoboLab-120 leaderboard, ahead of every open-weight VLA including π0.5 and GR00T N1.6
- Competitive success rate on real-world Franka tasks against state-of-the-art policy models
- Real-time control of a bimanual YAM robot and a single Franka arm on contact-rich tasks (placing a mug on a tray, boxing items, clearing bottles, closing drawers)
Fast adaptation: Adapts to new robots with under 30 minutes of data — all 15 simulated trials succeeded on Noble Machines' robot; competitive on Google RT-1 SimplerEnv; fine-tunable on the low-cost SO-101 for real-world tasks. Code, weights and integrations are open, with a technical report covering architecture and training stages.
Related event: Runway Open-Sources Praxis-1 World Action Model, Tops RoboLab-120(10 posts)→
More from Embodied
- Open-source dual-arm robot solves tangram puzzles with 90 demos, ~60% success — DanGrover · 2026-11-12
- Robotics has three regional playbooks: China volume, Europe inspection, US pilots — ingliguori · 2026-11-12
- 3,500 Laser-Armed AI Submarine Robots Patrol Norwegian Salmon Farms — philfung · 2026-11-12
- Model Autonomously Designs a Surgical-Robot-Style Robotic Arm CAD After 75 Minutes — _Stocko_ · 2026-11-12
- Dev releases local world model turning any image into interactive video, runs on RTX 3060 — lucidml_lover · 2026-11-12
- RoboRisk benchmark shows foundation models pose high harm rates in robotics; LEASH harness cuts risk — UoLiverpool · 2026-11-12