- Who
- vLLM
- What
- adds day-zero support for Runway’s Praxis-1 in vLLM-Omni, including offline inference and OpenPI / Positronic serving
- Quote
🎉 Day-0 support for @runwayml’s Praxis-1 in vLLM-Omni!
What happened
vLLM-Omni adds day-one support for Runway’s Praxis-1, a world action model that uses large-scale video pretraining for robot control and has open weights under Apache 2.0. The integration takes camera observations, robot state, and a task instruction to generate action chunks; it supports offline inference and OpenPI / Positronic serving, with session history and reset handling. The post says it was qualified on a single NVIDIA H100 80GB.
Why it matters
The post describes how vLLM-Omni connects Praxis-1 to robot inputs and action chunks, with serving and session handling details that clarify the integration’s scope.
Read next
See the full storyRunway releases Praxis-1 with Apache 2.0 open weights and vLLM-Omni support🎉 Day-0 support for @runwayml’s Praxis-1 in vLLM-Omni!
From watching the world to acting in it: Praxis-1 brings large-scale video pretraining to robot control, with open weights under Apache 2.0 and adaptation across robot embodiments.
vLLM-Omni takes camera observations, robot state and a task instruction to generate action chunks. The integration supports offline inference and OpenPI / Positronic serving, with session history and reset handling. Qualified on a single @NVIDIA H100 80GB.
Thanks to the Runway and @Inferact teams for bringing this together.
View the quoted post on XToday, we’re making Praxis-1 available to everyone. Our state-of-the-art world action model turns video pretraining into control for real robots, and it ships with open weights under Apache 2.0, ready to run and fine-tune on your own hardware.
The story so far
Runway has released Praxis-1 with open weights under the Apache 2.0 license, and vLLM-Omni added support for running it on a single NVIDIA H100 80GB GPU. Runway describes Praxis-1 as a world action model that uses video pretraining to control real robots, and says users can run and fine-tune it on their own hardware. The vLLM-Omni integration takes camera observations, robot state, and a task instruction to generate action chunks. It supports offline inference and OpenPI and Positronic serving, including session history and reset handling.
How the story developed · 2
View full event- 11/12 00:02 (UTC+8)
vLLM-Omni Adds Day-One Support for Runway’s Praxis-1 This report
- 11/12 00:02 (UTC+8)Runway Makes Praxis-1 Available with Apache 2.0 Open WeightsRunway · X