AlayaWorld: Long-Horizon and Playable Video World Generation

arXiv 2026 44.9 method, system

TLDR

AlayaWorld is an open-source framework for building interactive, playable video world models that generate long-horizon environments in real-time.

Reasoning

The paper presents a full-stack framework for generative world models with real-time interaction, but lacks explicit empirical evaluation details in the abstract. Its strengths include modular architecture and open-source release; weaknesses are unclear experimental validation.

Read-first score

Read-first score 44.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 55.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 29%

78.6

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=evaluation,metric

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Citation impact 18%

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. cited_by_count=0

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 469.

Keyword Scores

video world model

world model

generative world model

interactive world model

world dynamics prediction

world simulator

model-based reinforcement learning world model

Deep Analysis

Innovations

Full-stack open-source framework for interactive generative world building
Unified pipeline covering data preparation, model architecture, training, inference acceleration, and deployment
Real-time open-ended interaction enabling navigation, combat, spell casting, and monster summoning
Reproducible pipelines, reference implementations, evaluation tools, and documentation for future research

Methodology

The framework uses autoregressive video world models trained on gameplay recordings and real-world videos to synthesize future observations based on current world state and user interactions. It provides a modular and extensible architecture that integrates data preparation, model training, accelerated inference, and deployment.

Key Results

AlayaWorld enables real-time, playable world generation with diverse interactions such as combat and spell casting, but no quantitative metrics are reported in the abstract.

Resources

Paper