arXiv:2607. 00491v1 Announce Type: cross Abstract: Benchmarks for vision-language models (VLMs) mostly test observational spatial reasoning: models describe relations already visible in the input.
The paper introduces BedBench, a benchmark of thirteen procedurally generated tasks designed to test vision‑language models’ ability to recover and consistently use exact 3D object properties. Across twelve open‑weight and four frontier models, the study finds a stark gap: humans achieve 93.3 % accuracy on per‑color counting, while the best model scores only 7.2 %. Even targeted fine‑tuning can raise a model’s accuracy to 68.6 % but leaves its answers about the most common colors inconsistent with its own per‑color counts, highlighting that recovering a property does not guarantee its consistent use.
SceneTeract is a verification interface that separates semantic action understanding from physical feasibility in indoor 3D scenes. It decomposes activities into atomic actions and performs explicit geometric checks to determine executability, providing diagnostic traces for failures. The system reveals widespread functional and accessibility issues in synthetic scenes, shows that existing VLMs over‑predict action feasibility, and improves VLM performance through post‑training with verifier feedback, with benefits that generalize to real‑world scenes.
arXiv:2609.40325v1 Announce Type: new Abstract: As interactive 3D worlds are increasingly used to study intelligent behavior, it becomes important to develop efficient pipelines for identifying anoma...
The paper introduces a tool‑augmented framework that enhances a small Vision‑Language Model (Qwen3.5‑4B) with geometric tools—3D object detection, metric depth estimation, and deterministic solvers for distance, size, and bearing—to improve metric spatial reasoning. By moving metric computation from the model’s weights into explicit solvers, the approach achieves significant gains on ReVSI‑Bench tasks, notably increasing absolute distance accuracy from 0.46 to 0.74 MRA and relative direction accuracy from 25.9% to 73.4%. The modular design allows swapping in different detectors, enabling a clear separation between perception and reasoning errors, and the model can autonomously sequence the tools to match a scripted pipeline on most tasks.
arXiv:2609.09396v1 Announce Type: new Abstract: As Vision-Language Models (VLMs) advance toward physical deployment, the focus has remained on action-oriented Embodied AI evaluated on subject-centric...
Existing video benchmarks evaluate action recognition on consumer videos, egocentric recordings, or simulated industrial environments. They do not test vision-language models under the visual and procedural conditions of real industrial CCTV, where workers appear as distant figures amid dust, steam, low light, glare, occlusion, and overlapping activities.
AgentVQA is an offline benchmark designed to evaluate vision‑language models on realistic agentic scenarios across five domains: Web Agents, Robotics, Egocentric Videos, Games, and Spatial Planning. It comprises 1,000 carefully selected real‑world examples drawn from 23 source datasets, enabling reliable comparisons without costly online simulations. The benchmark reveals that current models achieve only 58.5% accuracy, with most errors stemming from difficulties in grounding and tracking evolving visual states rather than from task misunderstanding or reasoning.
The paper introduces the Separable Law, a framework that predicts how vision‑language model performance varies with language backbone size and visual token count. By fitting this law to 26 InternVL and QwenVL models across a range of backbone sizes (1B–72B) and image resolutions (224–8K pixels), the authors show that some question types scale predictably with model capacity while others do not. The law also provides a closed‑form rule for allocating compute between backbone size and visual tokens, helping to choose near‑optimal model and image sizes under a fixed budget.
PRISM-BENCH is a new benchmark comprising 1,000 expert‑curated vision‑language questions across five under‑represented reasoning categories, each paired with an idealized reasoning trace validated by a second annotator. It introduces diagnostics beyond final‑answer accuracy: a perception probe that attributes about 85% of frontier‑model failures to misperception, and trace alignment that predicts answer correctness with a macro AUROC of 0.66. The benchmark also presents CORE, an efficiency‑aware metric that reorders 24.2% of model pairs relative to accuracy while remaining uncorrelated with response length, revealing that recent gains are driven by recognition rather than spatial grounding.
The paper introduces a new benchmark for vision‑language models that tests their ability to decide whether to answer a physics question immediately or to request additional experimental evidence. Each problem presents one measurement image and four possible physical worlds defined by two masses and two values of another property; the model must either stop and answer or choose the cheapest experiment that resolves the question. Across six open models and 144 parameter sets, the models almost always repeat the same action even when the optimal choice changes, and only a single model gets both decisions correct on 5.9% of cases.
arXiv:2605. 05407v2 Announce Type: replace Abstract: Scaling LLM-based embodied agents from text-only environments to complex multimodal settings remains a major challenge.