dailyai.report

An international research team launched OpenWorldLib to standardize the definition of world models. The framework explicitly excludes text-to-video generators like Sora from this category. This distinction separates predictive physical simulations from mere visual synthesis. Practitioners can now use these criteria to evaluate whether a model actually simulates environment dynamics or just mimics pixels.