mlsanityData to 14 Oct 2026

MultiWorldBench: Do Independently Controlled Views Describe One Shared World?

arXiv 2610.11723 ↗First posted 8 Oct 2026cs.AI cs.CVcited by 0

On the map: World models → Video-generation world models and simulators

Abstract (arXiv)

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.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Video-generation world models and simulators, one of 11 directions in World models. Builds on 10 of its 14 references that are in the map. First posted on arXiv 8 Oct 2026.

Builds on

10 of 14 references.

Closest ancestors · 8

Same direction, newest first

  1. 2026Multiplayer Interactive World Models with Representation Autoencoders6 Jul
  2. 2026Gamma-World: Generative Multi-Agent World Modeling Beyond Two Players27 May
  3. 2026WBench: A Comprehensive Multi-turn Benchmark for Interactive Video World Model Evaluation25 May
  4. 2026WorldMark: A Unified Benchmark Suite for Interactive Video World Models23 Apr
  5. 2026Solaris: Building a Multiplayer Video World Model in Minecraft25 Feb
  6. 2026MIND: Benchmarking Memory Consistency and Action Control in World Models8 Feb
  7. 2025MineWorld: a Real-Time and Open-Source Interactive World Model on Minecraft11 Apr
  8. 2024Genie: Generative Interactive Environments23 Feb

Other references · 2

  1. 2018World Modelscited by 1,513
  2. 2025Matrix-Game: Interactive World Foundation Modelcited by 19

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026WorldBench: Evaluating LLMs on Three.js Voxel World Generation
  2. 2026MultiWorld: Scalable Multi-Agent Multi-View Video World Models
  3. 2026MASS: Multiplayer World Models with Authoritative Shared State
  4. 2026PAWA: Panoramic Action-conditioned World Modeling for Multiple Agents
  5. 2026World Editing: Intervening on Executable Worlds at Increasing Depth
  6. 2026PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives
  7. 2026MineExplorer: Evaluating Open-World Exploration of MLLM Agents in Minecraft
  8. 2026HappyWorld-Bench