{
  "id": 5472092,
  "title": "GraFT: A Training-Free Framework for Spatial Reasoning in Multimodal Large Language Models via 3D Scene Graphs",
  "url": "https://urgent.news/2026/09/03/graft-a-training-free-framework-for-spatial-reasoning-in-multimodal",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-03T14:11:56.000Z",
  "source": {
    "name": "arXiv cs.AI",
    "slug": "arxiv-cs-ai",
    "url": "https://arxiv.org/abs/2609.03892v1"
  },
  "original_language": "en",
  "account": null,
  "summary": "3D spatial reasoning underpins understanding and acting in the physical world, yet it remains unreliable in current multimodal large language models (MLLMs). These models falter at precise geometric measurement, at transforming between egocentric and allocentric viewpoints, and at grounding fine-grained appearance. The most common remedies fine-tune the model on large-scale curated…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "arXiv cs.AI",
        "title": "Large Language Models (LLMs) for Telecom Root Cause Analysis (RCA): A Structured Reasoning Framework for Evidence-Grounded Diagnosis",
        "url": "https://urgent.news/2026/09/02/large-language-models-llms-for-telecom-root-cause-analysis-rca-a",
        "published": "2026-09-02T16:43:22.000Z"
      }
    ]
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}