# WorldOracle benchmark finds world models often mispredict object motion in robot tasks

> Original title: WorldOracle Tests Whether World Models Can Reliably Predict Real-World Executions

- Source: The Robot Report
- Published: 2026-10-28T13:44:31.000Z
- Bad Influence: https://badinfluence.news/items/i5nysvt15nzetc740y6pk1orl
- Original: https://www.therobotreport.com/worldoracle-tests-whether-world-models-can-reliably-predict-real-world-executions/

## Summary

Princeton, UT Austin, University of Toronto, NVIDIA and AgiBot introduce WorldOracle, a benchmark that scores action-conditioned world models against real robot executions. Across seven models, including DreamDojo, Cosmos 3, IRASim and GE-Sim 2.0, arm motion was predicted relatively well while object positions and movements often deviated from reality. Scores also fell over complete operation sequences, with DreamDojo dropping from 50.56 to 27.08 and Cosmos 3 from 53.99 to 26.60, and the platform opened to developers on October 16.