Introducing V-JEPA 2 world model and new physical reasoning benchmarks

https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/

1.2B parameters; IntPhys 2, MVPBench, CausalVQA released; humans 85-95% on the benchmarks; current video models near chance on IntPhys 2; single time-scale limitation.

Physical-world foundation models