Runway Releases Praxis-1, a 27B MoE Video-Action Diffusion Transformer
Joint video-and-action prediction suggests Runway is steering its research toward world models that could serve robotics and simulation, not just media generation. Developers and researchers working on embodied AI or interactive video should read the technical report for architectural details unavailable in most commercial releases. It offers a direct comparison point against open world-model efforts that typically disclose far less about training design.
Runway announced Praxis-1, a 27B-parameter mixture-of-experts diffusion transformer that predicts video and actions together, and published a technical report detailing the model's architecture, training stages and results. The report is available at runway.com/research/praxis.
The model combines video prediction with action prediction in a single unified framework, built on a mixture-of-experts diffusion transformer architecture. This means that instead of generating video frames alone, Praxis-1 models the actions that drive what happens in the scene, a capability relevant to world models and embodied AI research. The technical report outlines how the training pipeline was staged to reach this joint prediction behaviour.
Praxis-1 positions Runway alongside other labs pursuing unified video-and-action world models, a direction that connects generative video work with robotics and simulation use cases. For teams tracking world-model development, the published report offers a direct look at how a production video company structures large-scale diffusion training.
Praxis-1 is a 27B-parameter mixture-of-experts diffusion transformer that predicts video and actions together. We detail the model's architecture, training stages and results in our technical report. runway.com/research/praxi…📊 1 ⚡
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Runway, Praxis-1, video generation, world models.
Source attribution: curated from Runway ML (@runwayml).
