Back to AI News
arxiv

MentalGrid Benchmark and SpatialWalker Model Target Dynamic Spatial Simulation in MLLMs

Researchers introduce MentalGrid, a benchmark for testing whether multimodal large language models can simulate evolving spatial states without intermediate visual feedback. The evaluation finds existing models struggle, with simple scenarios limited by perception and complex ones by multi-step reasoning. The authors propose a three-stage curriculum learning framework using 16k specialized samples, producing SpatialWalker, which improves 33.8% on average over its base model and outperforms Gemini-3-Flash on spatial simulation. SpatialWalker also shows a 4.9% average gain on out-of-domain benchmarks.

AI Brief

It provides a concrete new benchmark and training approach for a known weakness in multimodal models, with measured results useful to AI researchers.

arXiv:2611.22541v1 Announce Type: cross Abstract: Spatial reasoning is a fundamental component of world understanding for Multimodal Large Language Models (MLLMs). While existing research has advanced the perception of spatial relations in static scenes, the domain of dynamic spatial simulation remains relatively unexplored. To bridge this gap, we introduce MentalGrid, a benchmark focused on spatial reasoning tasks that require models to simulate evolving spatial states without intermediate visual feedback. Extensive evaluation on our benchmark reveals a significant deficiency in the spatial simulation capabilities of existing models. Specifically, we find that performance is constrained by perceptual limitations in simple simulation scenarios and hindered by inadequate multi-step reasoning as complexity increases. To address these challenges, we propose a three-stage curriculum learning framework utilizing 16k specialized spatial reasoning samples. This approach yields SpatialWalker, which achieves a 33.8% average improvement over its base model, surpassing proprietary models like Gemini-3-Flash in spatial simulation. Furthermore, SpatialWalker demonstrates robust generalization, maintaining a 4.9% average gain on out-of-domain benchmarks, demonstrating the pivotal role of spatial simulation as a fundamental building block for broader spatial intelligence.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!