{"slug": "code-as-worlds-agentic-discovery-of-executable-world-representations-for", "title": "Code as Worlds: Agentic Discovery of Executable World Representations for Physical Reasoning", "summary": "Researchers introduced Code-as-World, a paradigm that represents physical worlds as executable code, enabling compact and controllable abstractions for physical reasoning. The agentic discovery loop proposes, executes, renders, verifies, and refines world hypotheses from multimodal observations. Their model, Code-as-World-VL, achieved state-of-the-art performance on QuantiPhy, surpassing leading proprietary models.", "body_md": "arXiv:2608.27549v1 Announce Type: new\nAbstract: Physical understanding and reasoning depend on forming compact and generalizable representations of the world. While modern vision-language models can recognize and explain diverse physical events, they often lack explicit representations of the underlying mechanisms-such as object states, physical parameters, and governing dynamics-needed for reliably reasoning how the world evolves and responds to interventions. In this work, we introduce Code-as-World, a paradigm that represents physical worlds through executable world representations. By expressing physical composition, dynamic evolution, and visual appearance as executable code, Code-as-World provides a compact, quantitatively grounded, and controllable abstraction of the physical world. To construct such representations from multimodal observations, such as natural-language descriptions or real-world videos, we develop an agentic discovery loop inspired by abductive reasoning, where an agent proposes, executes, renders, verifies, and iteratively refines executable world hypotheses. As a concrete application, we use verified executable worlds to provide scalable physical supervision for training vision-language models on quantitative physical reasoning. Experiments show that Code-as-World-VL achieves state-of-the-art performance on QuantiPhy and surpasses leading proprietary models, highlighting the potential of executable world representations as a scalable foundation for physical intelligence.", "url": "https://wpnews.pro/news/code-as-worlds-agentic-discovery-of-executable-world-representations-for", "canonical_source": "https://arxiv.org/abs/2608.27549", "published_at": "2026-08-31 04:00:00+00:00", "updated_at": "2026-08-31 04:22:40.798867+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "machine-learning"], "entities": ["Code-as-World", "Code-as-World-VL", "QuantiPhy"], "alternates": {"html": "https://wpnews.pro/news/code-as-worlds-agentic-discovery-of-executable-world-representations-for", "markdown": "https://wpnews.pro/news/code-as-worlds-agentic-discovery-of-executable-world-representations-for.md", "text": "https://wpnews.pro/news/code-as-worlds-agentic-discovery-of-executable-world-representations-for.txt", "jsonld": "https://wpnews.pro/news/code-as-worlds-agentic-discovery-of-executable-world-representations-for.jsonld"}}