Population-Scalable Multi-Agent World Modeling

2026-08-09

AI Summary

This work addresses the limited scalability of existing world models in multi-agent environments, which typically rely on a fixed number of agents during both training and inference. To overcome this bottleneck, the authors propose the Khora framework, which decouples world state evolution from visual rendering and introduces a unified query-based rendering mechanism independent of agent count. By enforcing cross-view consistency over a shared world state—rather than through dense interactions among observation streams—Khora enables real-time interaction with an arbitrary number of agents during inference. The approach maintains high-fidelity visual generation and multi-agent consistency even when evaluated on unseen numbers of agents, thereby realizing a scalable open-world simulation system.

Abstract

World models have recently achieved impressive progress in visual prediction and interactive generation, but extending them to multi-agent environments introduces a fundamental scalability challenge. Existing methods generally assume a fixed number of agents during training and inference, which ties the model to a pre-determined agent population and limits inference-time scalability. Our key insight is that cross-view consistency should arise from a shared world state whose evolution does not assume a predefined number of agents, while agent-specific observations should be generated by querying this state through a unified rendering interface. Based on this insight, we propose Khora, a scalable multi-agent world model that supports inference-time expansion to arbitrary numbers of agents without retraining. Our framework decouples world-state evolution from visual rendering and introduces a population-agnostic rendering mechanism for incorporating other agent information. This design maintains cross-view consistency through the shared world state rather than through dense interactions among observation streams inside the expensive video generator, enabling approximately linear practical scaling with the number of queried views. Qualitative experiments demonstrate that our approach generalizes to unseen numbers of agents while maintaining visual quality and multi-agent consistency. We further implement a real-time interactive system to demonstrate scalable open-world simulation.

Problem

• multi-agent
• world modeling
• scalability
• population-agnostic
• cross-view consistency

Innovation

• scalable multi-agent world modeling
• population-agnostic rendering
• shared world state
• cross-view consistency
• inference-time expansion