MultiWorldBench is a diagnostic Minecraft benchmark containing 495 case configurations across seven task suites and ten capabilities: independent control, cross‑view motion, shared‑state synchronization, persistence, structural reasoning, concurrent interaction, and delayed revisit.
We evaluate three generated systems—Solaris, Gamma‑World, and MineWorld—using Engine GT as a reference. Gamma‑World achieves the highest average across the ten capabilities (21.39), followed by Solaris (20.88) and MineWorld (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 succeed in spatial reasoning or building‑identity preservation.
Human preference experiments 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 indicate that plausible individual views do not yet constitute a coherent multiplayer world.
Review