{
  "abstract": "Understanding a chemical reaction requires mapping a symbolic reactant-product description onto the three-dimensional pathway through which atoms rearrange, yet chemistry benchmarks for large language models (LLMs) largely probe factual knowledge and text-based reasoning. Here we introduce TSBench, a benchmark in which an LLM agent uses structure-editing tools to construct three-dimensional transition-state (TS) guesses verified by an automated quantum-chemical pipeline, yielding a physics-grounded pass/fail verdict. Across 546 evaluations of seven frontier LLMs on 78 elementary reactions, the aggregate success rate rose from 50.4% to 66.8% under diagnosis-driven revision; the best models approached 90% on the simplest reactions, yet performance dropped sharply with mechanistic complexity. Most failed attempts produced locally plausible saddle points whose reaction paths led to the wrong reactant-product pair, revealing local geometric intuition without a robust grasp of the global reaction coordinate. TSBench establishes a mechanism-level yardstick for LLM agents in mechanism-sensitive tasks such as synthesis planning and autonomous experimentation.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2609.08503",
  "author_total": 2,
  "authors": [
    {
      "name": "Xiaohu Xu"
    },
    {
      "name": "Tong Zhu"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/6dnys0wmzk9fh.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2609.08503",
  "fulltext_url": null,
  "id": "6dnys0wmzk9fh",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2609.08503",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "physics.chem-ph"
    ],
    [
      "openalex",
      "2505|Machine Learning in Materials Science"
    ]
  ],
  "openalex": "W7212031430",
  "pdf_url": "https://arxiv.org/pdf/2609.08503",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-08",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/6dnys0wmzk9fh.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:19:52Z",
      "native_id": "2609.08503",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2609.08503"
    }
  ],
  "subjects": [
    {
      "field": "chemistry",
      "origin": "arxiv:physics.chem-ph",
      "subfield": "physical"
    }
  ],
  "title": "TSBench: A physics-grounded benchmark for evaluating LLM understanding of chemical reaction mechanisms",
  "title_text": "TSBench: A physics-grounded benchmark for evaluating LLM understanding of chemical reaction mechanisms",
  "type": "preprint",
  "updated_at": "2026-10-30T12:19:52Z",
  "url": "https://research.draconitic.com/paper/6dnys0wmzk9fh",
  "venue": "arXiv",
  "versions": []
}