Observed arrival · 2026-10-14
Crusader Bench puts AI agents in medieval town-planning trials
Crusader Bench runs AI models through Stronghold Crusader maps and scores the village economies they build against idle and human-played baselines.
- For
- AI evaluation and strategy-game enthusiasts
- Worth noticing
- Cross-map comparisons include only models with full-budget scores on all three maps, then average their map scores.
Field notes
The score assigns 80% weight to net worth, with population and popularity contributing 10% each; charts also compare performance against game cost and population growth. Cross-map comparisons include only models with full-budget scores on all three maps, using the mean of their map scores. The dated changelog documents version-specific comparability and concrete run issues, including an armory-storage blind spot that left 14 workshops queued behind a full armory.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue