Spatial-Gym benchmark shows the best tested model solves only 16% of pathfinding tasks versus 98% for humans, with step-by-step and backtracking formats producing mixed effects across model strengths.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Mind the Gap Between Spatial Reasoning and Acting! Step-by-Step Evaluation of Agents With Spatial-Gym
Spatial-Gym benchmark shows the best tested model solves only 16% of pathfinding tasks versus 98% for humans, with step-by-step and backtracking formats producing mixed effects across model strengths.