NVIDIA and a joint research team from the Massachusetts Institute of Technology and the University of Oxford have introduced Physis-Lang, a framework designed to strengthen physical reasoning in video generation models through language. According to the Techub summary, the system adds physical reasoning fields to video descriptions and generates physical negative prompts to reduce implausible outputs during inference.
On the Physics-IQ Verified leaderboard dated Sept. 29, 2026, Cosmos3-Super with Physis-Lang ranked first with a score of 48.2, while Cosmos3-Nano placed second with 43.3. The report also said the framework outperformed Google’s Veo 3.1 across multiple physics benchmark tests.
NVIDIA AI later described Physis-Lang on X as an open-source, self-evolving framework. The post said it improves video descriptions to fine-tune models so generated videos better follow physical laws. The item was credited to MarkTechPost.
NVIDIA, together with research teams from the Massachusetts Institute of Technology and the University of Oxford, has introduced Physis-Lang, a framework aimed at improving physical reasoning in video generation models through language.
According to the Techub item, the framework adds physical reasoning fields to video descriptions and generates physical negative prompts to help avoid physically implausible results at the inference stage.
Leaderboard results
On the Physics-IQ Verified leaderboard dated Sept. 29, 2026, Cosmos3-Super equipped with Physis-Lang ranked first with a score of 48.2. Cosmos3-Nano placed second with 43.3.
The report added that the framework outperformed Google’s Veo 3.1 model on multiple physics benchmark tests.
Open-source framework details
NVIDIA AI also introduced the open-source, self-evolving framework in a post on X. The company said the system fine-tunes models by improving video descriptions, with the goal of making generated videos align more closely with physical laws.
The item credited MarkTechPost as the source.
This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan. Disclaimer:
The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.
Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.
