Driving · 2026-10-09
MultiWorldBench: Do Independently Controlled Views Describe One Shared World?
Zhangbo Xu, Ruoxi Zhang, Rui Hu, Yisong Wang
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
The authors introduce MultiWorldBench, a Minecraft diagnostic benchmark (495 cases) for multiplayer world models: two independently controlled views should reflect one shared, persistent world. They evaluate Solaris, Gamma-World, and MineWorld vs Engine GT. Gamma-World ranks highest among generated systems (21.39 ten-capability average), but all generated systems score very low on state persistence and structural consistency, and none succeeds on spatial reasoning or building-identity preservation.
Why it matters
Multiplayer world models must ensure that multiple independently controlled views remain consistent with one shared world (control isolation, shared-state synchronization, persistence, identity preservation). MultiWorldBench provides capability-specific, cross-view diagnostics for falsifiable consistency checks, and the results show that plausible individual views do not yet guarantee coherent shared worlds.
Method
- Introduce MultiWorldBench: 495 Minecraft case configurations across seven task suites and ten capabilities (independent control, cross-view motion, shared-state sync/persistence, structural/spatial reasoning, concurrent interaction, delayed revisit).
- Evaluate Solaris, Gamma-World, MineWorld, using Engine GT as reference.
- Use visibility-conditioned paired observations and action/trajectory + visual judgment metrics, plus human preference win-ratio alignment (Spearman ρ mean 0.96).
Limitation
Our evaluation is limited to Minecraft; its discrete voxel-based visual style is substantially different from continuous photorealistic real-world dynamics, and the authors note future evaluations should include more cases, environments, and action sequences to test generalization.
Abstract (from 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.
Related papers
- World Editing: Intervening on Executable Worlds at Increasing Depth
- PAWA: Panoramic Action-conditioned World Modeling for Multiple Agents
- ROWBench: Do Video Models Render What the Program Specifies?
- ExeWorldBench: Benchmarking Code-Based Executable 3D World Modeling
- OverSeer: Multi-Agent World Models Beyond What Agents See
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-08 · Indexed by Pub-AI: 2026-10-09 UTC
MultiWorldBench: Do Independently Controlled Views Describe One Shared World?
Preprint posted: 2026-10-08 · Original record ↗
- arxiv
- 2610.11723
Explicit source versions (1)
- v4 · 2026-10-20
- Preprint available
Source date: 2026-10-08 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (2)
- Official evidence ↗ · OAI arXiv created; not header datestamp (official metadata)
- Official evidence ↗ · atom:published (official metadata)
- Explicit source version detected · v4
Source date: 2026-10-20 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:id explicit version (official metadata)
Abstract · arxiv
Multiplayer world models must ensure that independently controlled views remain consistent with a single shared and persistent world. We introduce MultiWorldBench, a diagnostic Minecraft benchmark comprising 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. Using Engine GT as a reference, we evaluate two multiplayer world models, Solaris (MP) and Gamma-World, together with a single-player generation variant, Solaris (SP). Among the generated systems, Gamma-World achieves the highest ten-capability average at 21.39, followed by Solaris (MP) at 20.88 and Solaris (SP) at 10.06, while Engine GT reaches 91.69. Gamma-World performs better on several control, shared-state, and revisit capabilities, whereas Solaris (MP) 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 pairwise evaluation yields the same overall system ranking and shows strong alignment with the automatic evaluation, with a mean dimension-level Spearman correlation of 0.95. These results show that current multiplayer world models still exhibit a substantial gap relative to Engine GT in maintaining a coherent shared world.