MindTopo is a benchmark that tests foundation models on topological reasoning, covering five cognitive properties—continuity, separation, order, enclosure, and knots—across two cognitive levels: reasoning and planning. It contains 11,030 instances from 13 procedurally generated task types, and evaluates 14 multimodal large language models, including agent configurations with image and video generation. Results show that models perform better on reasoning than planning, and even the best model lags far behind human performance, with fine‑tuning and reinforcement learning improving reasoning more than planning.
A path, a fence, a knot. MindTopo sets a new benchmark for testing how AI understands topological relationships and highlights new opportunities to strengthen spatial reasoning and planning.
arXiv:2603. 10384v3 Announce Type: replace Abstract: Evaluating LLM reliability via scalar probabilities often fails to capture the structural dynamics of reasoning.
Think3D introduces a framework that endows Vision‑Language Models with interactive 3D chain‑of‑thought reasoning by integrating 3D manipulation tools for active spatial exploration. The approach improves performance on benchmarks such as BLINK Multi‑view, MindCube‑1K, and VSI‑Bench‑Tiny for proprietary models like GPT‑4.1 and Gemini 2.5 Pro, and a reinforcement‑learning variant, Think3D‑RL, enables open‑weight models such as Qwen3‑VL‑4B to autonomously learn effective 3D exploration strategies, yielding tool‑use patterns comparable to stronger models and turning a performance drop on MindCube‑1K into a substantial improvement.
arXiv:2606. 08952v1 Announce Type: new Abstract: Multimodal Foundation Models (MFMs) have made substantial progress, yet remain fragile in spatial reasoning over the physical world.
The paper introduces NavMCP, a scaffolding framework that couples vision‑language models (VLMs) with navigation foundation models (NFMs) to enable long‑horizon physical‑world agents. NavMCP orchestrates three communication channels—intent, observation, and memory—to allow the VLM to decide what evidence to seek and the NFM to ground semantic sub‑goals into closed‑loop navigation, without retraining either model. The approach achieves state‑of‑the‑art results on several embodied question‑answering benchmarks and significantly outperforms episodic interfaces on the Unitree Go2 robot as task horizons lengthen.
arXiv:2503. 19990v4 Announce Type: replace Abstract: Many real-world applications of spatial intelligence, such as robotic control, autonomous driving, and automated assembly, require spatial reasoning across multiple sequential steps.
arXiv:2601.13247v2 Announce Type: replace-cross Abstract: Current Large Language Models (LLMs) exhibit a critical modal disconnect: they possess vast semantic knowledge but lack the procedural ground...
PlanCraft introduces a progressive approach to 3D residential scene generation that mirrors how architects design: starting with rough sketches and refining them over time. It leverages a large dataset of real floor plans to train a SketchPlan module that generates partial sketches at various completion levels, a PlanCraft‑Diff module that sharpens these sketches into precise vector floor plans, and a PlanCraft‑Agent that furnishes rooms within the established spatial contract. The method outperforms existing 2D and 3D baselines, achieving a 61.1% lower FID and a 15‑point lead in expert‑rated spatial rationality, even with only 25% sketch completion.
arXiv:2606. 12830v1 Announce Type: cross Abstract: While recent vision-language models (VLMs) demonstrate strong multimodal understanding, they remain limited in spatial reasoning tasks that require active evidence acquisition and multi-step visual interaction.
arXiv:2608. 04575v1 Announce Type: cross Abstract: Reliable physical reasoning from video requires understanding how objects move, interact, and respond to interventions.
arXiv:2607. 22732v1 Announce Type: new Abstract: LLM-based game agents often perform poorly on more complex tasks.
The paper introduces FactoSR, a factorized reinforcement learning framework designed to improve spatial reasoning in Vision‑Language Models by addressing a dimensional mismatch between 2D visual inputs and the 3D+temporal nature of the physical world. FactoSR decomposes the reasoning task into three orthogonal geometric sub‑objectives—planar correspondence (XY), depth consistency (Z), and temporal reversibility (T)—and optimizes these constraints within a unified policy learning mechanism. Experiments on multi‑view and video benchmarks show that this decomposition yields significant performance gains, achieving a 5.9% improvement on VSI‑Bench and 4.5% on All‑Angles‑Bench.