Runway unveils Praxis-1, a 27B MoE diffusion Transformer
11/20/2026 — 11/20, 04:18·1 sources (all)·3 reports (all)
Key takeaways
- Runway has unveiled Praxis-1, a 27B-parameter mixture-of-experts diffusion Transformer that predicts video and actions jointly. The company…
- release timing and availability were not disclosed.
Background & analysis
On November 20, 2026, Runway announced Praxis-1 through its official X account (@runwayml). The company described Praxis-1 as a 27B-parameter mixture-of-experts diffusion transformer that predicts video and actions together, and said the technical report lays out the model's architecture, training stages, and results.
The same day, Runway shared additional claims about Praxis-1's task learning and execution performance on X. According to the company, with less than 30 minutes of teleoperated demonstration data, Praxis-1 succeeded in all 15 simulated trials on Noble Machines' robot, showing that it can pick up new tasks from very little data. On throughput and efficiency, Runway stated that when driving a robotic arm on the same station and the same set of tasks, Praxis-1 completed tasks 15.6x faster than GPT-6 Astra while costing 51x less per completed task. These comparisons were published by Runway itself, and the benchmark setup was not independently verified in the available coverage.
At this point, the public record consists of the architecture description and the performance and cost figures released by Runway. Details on the technical report's release timing and any external evaluation of the model have not been covered in the current reports.
AI-generated from 3 reports · updated 2 days ago
Latest turnRunway has detailed Praxis-1, a 27B-parameter mixture-of-experts diffusion transformer that predicts video and actions together. The company says its technical report covers the model's architecture, training stages and results.
Related AI tools
- Heat index
- 25
- All sources
- 1
- All reports
- 3
- First seen
- 2 days ago
Reports on this story headlines open the original
Runway has detailed Praxis-1, a 27B-parameter mixture-of-experts diffusion transformer that predicts video and actions together. The company says its technical report covers the model's architecture, training stages and results.
X:Runway (@runwayml)AI score 60Runway says Praxis-1 can pick up new tasks from very little data, succeeding in all 15 simulated trials on Noble Machines' robot after training on under 30 minutes of teleoperated demonstrations.
X:Runway (@runwayml)AI score 57Runway claims that on the same station and tasks, Praxis-1 drove a robot arm to completion 15.6x faster than GPT-6 Astra, at 51x lower cost per completed task. The comparison comes from Runway itself.
X:Runway (@runwayml)AI score 55
Other stories people are talking about
- Heat index 37Runway launches Agent to turn short conversations into video1 sources
- Heat index 42Runway says nearly all merged PRs are written by agents1 sources
- Heat index 290RisingGPT-6.2 Astra and Sol now generally available, with deployment comparisons5 sources
- Heat index 77Microsoft rolls out Copilot integrations and unveils MAI models at Ignite 20262 sources
- Heat index 62OpenAI report says its models rewrote AMD GPU RTL, moves into manufacturing1 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

