{
  "id": 10589382,
  "title": "The Decomposition Tax: LLM Pipelines Lose Up to 40 Accuracy Points at Their Own Interfaces",
  "url": "https://urgent.news/2026/09/26/the-decomposition-tax-llm-pipelines-lose-up-to-40-accuracy-points-at",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-26T17:49:27.000Z",
  "source": {
    "name": "arXiv cs.AI",
    "slug": "arxiv-cs-ai",
    "url": "https://arxiv.org/abs/2609.32825v1"
  },
  "original_language": "en",
  "account": null,
  "summary": "A four-stage LLM pipeline gives up as much as 40.5 accuracy points at its own interfaces (gemma-3-12B on MATH-500, Holm-corrected p = 1.66e-19; the largest tax in the primary family). We hold model, problem, stages, stage prompts and completion budget fixed, vary only whether each stage can still see the original problem, and call the accuracy difference the decomposition tax. Across 21…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "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."
}