{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-11-06-016",
    "source_story_id": "tmp-brief-interactive-visual-gym",
    "edition_id": "mp-2026-11-06-morning-0119",
    "edition_url": "https://themachinepress.com/edition/2026-11-06",
    "position": 18,
    "story_type": "ticker",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "The Visual Assistant Enters a Closed Loop",
    "slug": "the-visual-assistant-enters-a-closed-loop",
    "dek": "IVG-Bench simulates users and executable tasks across cooking, household manipulation and mobile interfaces.",
    "summary": "IVG-Bench simulates users and executable tasks across cooking, household manipulation and mobile interfaces.",
    "body_text": "Across 315 scenarios, the best tested model finishes 53.4 percent within the time limit. A human study finds 78.5 percent agreement with the interaction-quality metric, and a small-model training experiment improves task success and quality in the cooking environment.",
    "why_it_matters": "IVG-Bench simulates users and executable tasks across cooking, household manipulation and mobile interfaces.",
    "limitations": [],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-06-016/the-visual-assistant-enters-a-closed-loop",
    "json_url": "https://themachinepress.com/story/mp-2026-11-06-016.json",
    "first_published_at": "2026-11-06T09:00:00.000-05:00",
    "modified_at": "2026-11-06T09:00:00.000-05:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-11-06-016-001",
        "text": "IVG-Bench simulates users and executable tasks across cooking, household manipulation and mobile interfaces.",
        "source_ids": [
          "source-2026-11-06-018"
        ],
        "qualification": null
      }
    ],
    "source_ids": [
      "source-2026-11-06-018"
    ],
    "tags": [
      "visual assistants",
      "simulation",
      "closed-loop evaluation"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-11-06-018",
      "title": "arXiv preprint 2611.08286, original v1",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2611.08286v1",
      "canonical_url": "https://arxiv.org/abs/2611.08286v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-11-05T11:50:31.000-05:00",
      "accessed_at": "2026-11-06T08:21:03.209-05:00",
      "supports_claim_ids": [
        "claim-mp-2026-11-06-016-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Visual Assistant Enters a Closed Loop",
    "publisher": "The Machine Press",
    "published_at": "2026-11-06T09:00:00.000-05:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-06-016/the-visual-assistant-enters-a-closed-loop"
  }
}