| title | Ψ-0.6: Adaptive Streams of Thought in a Unified Vision-Language World Model |
|---|---|
| authors | Khai Loong Aw, Klemen Kotar, Baihan Zhang, Yifan Liu, Lilian Chen, Atlas Kazemian, Rahul Venkatesh, Daniel Yamins |
| date | 2026-10-23 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2610.32177 |
| affiliation | Stanford AI |
| upvotes | 2 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| primer | World models |
| abstract | Humans reason about the physical world by flexibly combining different forms of inference as a problem unfolds. Achieving this flexibility is a major open challenge in AI: capabilities often remain separated across specialized models, limiting how they can be composed and jointly adapted to a task. We introduce Ψ-0.6, a unified vision-language world model that adaptively composes perception, language, geometry, and simulation within a shared reasoning process we call streams of thought. These streams interleave images, text, and explicit representations of geometry and motion, all of which the model can generate and interpret through a shared token interface. Ψ-0.6 surpasses leading open-weight models zero-shot on multi-view spatial reasoning, with further gains from reasoning through geometric intermediates. It also supports counterfactual simulation: generating and interpreting imagined scenes to determine, for example, whether moving an object would obstruct a pathway. Reinforcement learning from task feedback jointly optimizes the selection, generation, and use of intermediate representations, improving visual question-answering accuracy with gains that transfer to unseen benchmarks. Ψ-0.6 offers a path toward adaptive physical reasoning -- not only predicting the physical world but also choosing the representations through which to reason. |
cite
@misc{aw2026adaptive,
title={Ψ-0.6: Adaptive Streams of Thought in a Unified Vision-Language World Model},
author={Khai Loong Aw and Klemen Kotar and Baihan Zhang and Yifan Liu and Lilian Chen and Atlas Kazemian and Rahul Venkatesh and Daniel Yamins},
year={2026},
eprint={2610.32177},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.32177}
}