Monday, November 16, 2026

Top 5 This Week

Related Posts

Runway releases Praxis-1, a video model for piloting robots


Runway releases open source Praxis-1, a video model designed to perform robotic movements from images and text instructions. Internal benchmarks place this World Action Model at the top of its class on RoboLab-120, while detailing its limitations, data requirements and hardware requirements.

Known limits, explicit steps and still internal checks

Runway indicates that Praxis-1 sometimes confuses similar-looking objects and does not yet reliably plan complex action sequences autonomously. The model also struggles with vague instructions. For long tasks, it is necessary to explicitly formulate each step. The published results come from tests carried out by Runway and have not yet been independently verified. The company believes performance is promising in its internal evaluations.

Open source access and required hardware resources

Praxis-1 is offered with weight and code under the Apache 2.0 license, which allows commercial use. The template can be downloaded via Hugging Face after registering and agreeing to access terms, and the code is available on GitHub. Runway announces pre-adapted versions for different robots, in addition to the base model. The latter requires a GPU with 80 GB of memory. According to Runway estimates, a compressed version should run on an RTX 5090 with 32 GB of VRAM, but maximum memory usage has not yet been measured according to GitHub documentation.

Benchmarks: 46.2% on RoboLab-120 and 88 successes on real arm

On Nvidia’s RoboLab-120 benchmark, which evaluates 120 simulated sorting, grabbing, and stacking tasks, Praxis-1 achieves a 46.2% success rate. FLUX 3 Action, built on a similar approach, scores 42.9%, placing Praxis-1 at the top of the World Action Models category and fifth overall, behind two AI agents and two robotic models of different architectures. After an adjustment on DROID, a test on a real Franka arm, with ten tasks and ten attempts per task, resulted in 88 successes for Praxis-1 compared to 80 for FLUX 3 Action. Runway also evaluated the transfer to a mobile platform with a gripping arm not seen in training, resulting from Google’s work for RT-1. After approximately 95 hours of adjustment, the model achieved, according to Runway, 80% average success in simulation, on four types of tasks including gripping cans, opening and closing drawers, and placing an apple in a drawer. On this same benchmark, RT-1 obtains 52.4%.

The AI ​​brief that professionals read

Image, video, sound, generative design: the best every evening in 5 minutes. Free.

Included upon registration: our selection of the best AI guides & comparisons.

Choose your pace

Free · No spam · Unsubscribe in 1 click

Training data: general video, DROID and ever-present robot need

Praxis-1 is based on a general video model refined with videos of human actions and data from robots. This approach could reduce dependence on strictly robotic datasets, but still requires dedicated corpora. To adapt the Franka one-arm model, approximately 277 hours of the public game DROID were used, with scenes of grabbing and moving objects. For a double-arm robot, the training required between 1,400 and 2,600 hours of data. Runway says it has already integrated data from DROID and other systems at earlier stages. It is not established that the approach requires less total robotic data than models without general video pretraining.

Operation, positioning and testing partners

The model processes camera images, textual instructions, and position information to predict scene developments and calculate appropriate robot movements, adjusting its control as new camera frames arrive. Runway describes Praxis-1 as a World Action Model and mentions trials conducted since September with Standard Bots, Noble Machines and Ultra on various platforms, including a humanoid robot, an industrial arm and a dual-arm warehouse robot. According to Runway, the model can be adapted to new systems by additional training. Presented to partners at the end of September, Praxis-1 is now accompanied by a detailed technical report. Runway, known for its generative video models like Gen-4.5, positions this system for rapid execution of movements and envisions that it could eventually work alongside general models responsible for breaking down complex tasks.

Shredder Smith
Shredder Smith
Shredder Smith is the lead curator and digital persona behind topaitools4you.com, an AI directory dedicated to "shredding" through industry hype to identify high-utility software for everyday users. Smith positions himself as a blunt, no-nonsense reviewer who vets thousands of emerging applications to filter out overpriced "wrappers" in favor of tools that offer genuine ROI and practical productivity. The site serves as a watchdog for the AI gold rush, providing categorized rankings and transparent reviews designed to help small businesses and creators navigate the crowded tech landscape without wasting money on low-value tools.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Popular Articles