Video World Models · 2026-10-14
ExeWorldBench: Benchmarking Code-Based Executable 3D World Modeling
Xinge Peng, Zhen Fang, Ruibing Wang, Yongxuan Lv, Haodong Li, Guohui Zhang, Xiaoxiao Ma, Yiting Lu, Lin Chen, Zehui Che, Feng Zhao
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
ExeWorldBench is presented as a benchmark for executable 3D world modeling via code-based 3D world construction. It evaluates how models translate physical understanding into executable realization across three tasks: Mechanism Executability (visual-to-Blender program), Compositional Executability (multi-stage causal chains), and Intent-Driven Executable World Synthesis (component- or goal-constrained). The authors report substantial limitations as causal complexity and execution horizon increase.
Why it matters
The authors argue that existing benchmarks test physical reasoning, code/geometry construction, or behavioral correctness in isolation, leaving insufficient evaluation of the integrated capability needed for executable world modeling. ExeWorldBench targets this integration by evaluating code executability and physically grounded structural/mechanical/causal correctness across increasingly autonomous synthesis settings.
Method
- Define executable 3D world modeling as the integration of world understanding (structural/mechanical/causal principles) and executable realization (code-based executable system behavior).
- Create three progressively challenging tasks: T1 Mechanism Executability (image to Blender program with complete motion), T2 Compositional Executability (multi-stage causal chain), and T3 Intent-Driven Executable World Synthesis (component- or goal-constrained).
- Evaluate generated worlds with Opus 5 as a VLM judge using metrics for Visual Structure, Mechanical Motion (T1), Executable Causal Mechanism (T2/T3), and Goal Satisfaction (T3-2), plus code executability with repair.
Limitation
The paper text provided does not state a specific limitation.
Abstract (from arXiv)
Coding agents are increasingly evolving beyond reproducing what a world looks like toward constructing executable worlds that capture how a world works. This shift calls for executable world modeling, which integrates two core capabilities: world understanding, reasoning about the underlying structural, mechanical, and causal principles of a world, and executable realization, translating that understanding into a code-based system whose behavior emerges from physical interactions. However, existing benchmarks provide only fragmented evaluations centered on physical reasoning, visual fidelity, or behavioral correctness, leaving the integration of these capabilities insufficiently assessed. To bridge this gap, we introduce ExeWorldBench, to our knowledge the first benchmark to comprehensively evaluate world understanding and executable realization through code-based 3D world construction. ExeWorldBench comprises three progressively challenging tasks: Mechanism Executability, which evaluates the realization of individual mechanisms from visual observations; Compositional Executability, which extends to multi-stage systems with spatial, mechanical, and causal dependencies; and Intent-Driven Executable World Synthesis, which requires constructing executable worlds from high-level goals and physical requirements. Extensive experiments on frontier models reveal substantial limitations in executable world modeling, offering insights into the development of more physically grounded and causally coherent world construction systems.
Related papers
- Code2Worlds-2: Physically Plausible Agent Harness for Embodied World Simulation
- Video2World: Benchmarking Coding Agents for Interactive World Modeling from Embodied Videos
- World Editing: Intervening on Executable Worlds at Increasing Depth
- WorldCast: Exploring Intelligent Visual Generation in World State Prediction
- WorldStream: Streaming Interactive 3D Worlds with Hierarchical Planning and Persistent State
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-10 · Indexed by Pub-AI: 2026-10-14 UTC
ExeWorldBench: Benchmarking Code-Based Executable 3D World Modeling
Preprint posted: 2026-10-10 · Original record ↗
- arxiv
- 2610.15166
Explicit source versions (1)
- v1 · 2026-10-10
- Preprint available
Source date: 2026-10-10 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
Coding agents are increasingly evolving beyond reproducing what a world looks like toward constructing executable worlds that capture how a world works. This shift calls for executable world modeling, which integrates two core capabilities: world understanding, reasoning about the underlying structural, mechanical, and causal principles of a world, and executable realization, translating that understanding into a code-based system whose behavior emerges from physical interactions. However, existing benchmarks provide only fragmented evaluations centered on physical reasoning, visual fidelity, or behavioral correctness, leaving the integration of these capabilities insufficiently assessed. To bridge this gap, we introduce ExeWorldBench, to our knowledge the first benchmark to comprehensively evaluate world understanding and executable realization through code-based 3D world construction. ExeWorldBench comprises three progressively challenging tasks: Mechanism Executability, which evaluates the realization of individual mechanisms from visual observations; Compositional Executability, which extends to multi-stage systems with spatial, mechanical, and causal dependencies; and Intent-Driven Executable World Synthesis, which requires constructing executable worlds from high-level goals and physical requirements. Extensive experiments on frontier models reveal substantial limitations in executable world modeling, offering insights into the development of more physically grounded and causally coherent world construction systems.