MultiWorldBench：独立控制的视图能描述同一个共享世界吗？

View PDF HTML (experimental)

Abstract:Multiplayer world models must ensure that independently controlled views remain consistent with one shared and persistent world. We introduce MultiWorldBench, a diagnostic Minecraft benchmark containing 495 case configurations across seven task suites and ten capabilities, including independent control, cross-view motion, shared-state synchronization, persistence, structural reasoning, concurrent interaction, and delayed revisit. We evaluate Solaris, Gamma-World, and MineWorld, using Engine GT as a reference. Gamma-World achieves the highest ten-capability average among the generated systems at 21.39, followed by Solaris at 20.88 and MineWorld at 1.89, while Engine GT reaches 91.69. Gamma-World performs better on several control, shared-state, and revisit capabilities, whereas Solaris leads in cross-view motion and race-condition consistency. Nevertheless, all generated systems score at most 8.00 on state persistence and 1.33 on structural consistency, and none succeeds in spatial reasoning or building-identity preservation. Human preferences produce the same overall ranking and show strong alignment with the automatic evaluation, with a mean dimension-level Spearman correlation of 0.96. These results show that plausible individual views do not yet constitute a coherent multiplayer world.

Comments:	31 pages, 17 figures, 3 tables
Subjects:	Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Cite as:	arXiv:2610.11723 [cs.AI]
	(or arXiv:2610.11723v1 [cs.AI] for this version)
	https://doi.org/10.48550/arXiv.2610.11723 arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Rui Hu [view email]
[v1] Thu, 8 Oct 2026 11:24:34 UTC (24,940 KB)