{
  "id": "arxiv:2604.17817",
  "citation-key": "arxiv_2604_17817",
  "type": "article",
  "title": "Do LLMs Need to See Everything? A Benchmark and Study of Failures in LLM-driven Smartphone Automation using Screentext vs. Screenshots",
  "URL": "https://arxiv.org/abs/2604.17817v1",
  "custom": {
    "catalog_id": "arxiv:2604.17817",
    "catalog_kind": "paper",
    "catalog_review_status": "curated",
    "canonical_source_url": "https://arxiv.org/abs/2604.17817",
    "identifiers": {
      "arxiv": "2604.17817"
    },
    "first_published_at": "2026-04-20",
    "record_updated_at": "2026-04-20",
    "catalog_scope": "core",
    "arxiv_snapshot": {
      "abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1",
      "abstract_sha256": "176f4d68e2f5084bee1938ee1399978052cdce2d2250ef06c16ddbf49c977da1",
      "submitted_at": "2026-04-20T05:15:14Z",
      "updated_at": "2026-04-20T05:15:14Z",
      "version": 1,
      "versioned_url": "https://arxiv.org/abs/2604.17817v1"
    }
  },
  "note": "First published: 2026-04-20; Recorded arXiv version: v1; Recorded version updated: 2026-04-20T05:15:14Z.",
  "author": [
    {
      "literal": "Shiquan Zhang"
    },
    {
      "literal": "Tianyi Zhang"
    },
    {
      "literal": "Le Fang"
    },
    {
      "literal": "Simon D'Alfonso"
    },
    {
      "literal": "Hong Jia"
    },
    {
      "literal": "Vassilis Kostakos"
    }
  ],
  "issued": {
    "date-parts": [
      [
        2026,
        4,
        20
      ]
    ]
  },
  "archive": "arXiv",
  "archive_location": "2604.17817",
  "genre": "preprint",
  "version": "1"
}