Physics-IQ
A benchmark designed to evaluate the degree to which generative models understand physical laws. The test set covers physical phenomena including fluid dynamics, elastic collisions, gravitational fall, and rigid-body rotation, assessing causal reasoning correctness rather than visual quality alone.
Timeline (last 90 days)
The company officially claims that the Odyssey-3 Pro version achieved the highest reported score of 66.1 (best-of-8) in the video-to-video evaluation on Physics-IQ Verified.
Physics-IQ is a benchmark measuring generative models' understanding of physical laws, with a test set covering physical phenomena such as fluid dynamics, elastic collisions, gravitational falls, and rigid body rotation.
Physics-IQ measures the correctness of causal reasoning, which differs from traditional visual quality metrics such as FID that measure the perceptual realism of images.
Best-of-8 means taking the best result from eight generations; this score is self-reported by the vendor, and actual stable performance still requires third-party verification.
All Facts (4)
The company officially claims that the Odyssey-3 Pro version achieved the highest reported score of 66.1 (best-of-8) in the video-to-video evaluation on Physics-IQ Verified.
50%UnverifiedPhysics-IQ is a benchmark measuring generative models' understanding of physical laws, with a test set covering physical phenomena such as fluid dynamics, elastic collisions, gravitational falls, and rigid body rotation.
50%UnverifiedPhysics-IQ measures the correctness of causal reasoning, which differs from traditional visual quality metrics such as FID that measure the perceptual realism of images.
50%UnverifiedBest-of-8 means taking the best result from eight generations; this score is self-reported by the vendor, and actual stable performance still requires third-party verification.
50%