{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-15-010",
    "source_story_id": "tmp-story-geothinker-predictive-latents",
    "edition_id": "mp-2026-10-15-morning-0097",
    "edition_url": "https://themachinepress.com/edition/2026-10-15",
    "position": 10,
    "story_type": "dispatch",
    "section": "frontier-models",
    "editorial_classification": "editorial",
    "headline": "The Spatial Model Dismisses Its 3D Encoder at Runtime",
    "slug": "the-spatial-model-dismisses-its-3d-encoder-at-runtime",
    "dek": "GeoThinker-F learns predictive geometric and future-semantic latents from an offline expert, then removes that expert for inference.",
    "summary": "GeoThinker-F learns predictive geometric and future-semantic latents from an offline expert, then removes that expert for inference.",
    "body_text": "External 3D encoders provide useful geometry but add a second runtime path. GeoThinker-F first trains lightweight heads to predict geometry-expert features and future semantics while the main multimodal model is frozen. It then removes the expert and teaches the backbone to fuse its own predictions. The authors report up to a 3.6-times speedup, cutting latency from 1.45 seconds to 0.40 seconds while retaining competitive performance across their spatial benchmarks. The efficiency claim is tied to the evaluated stack, but the handoff from teacher to internal representation is explicit.",
    "why_it_matters": "GeoThinker-F learns predictive geometric and future-semantic latents from an offline expert, then removes that expert for inference.",
    "limitations": [],
    "importance": 7,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-15-010/the-spatial-model-dismisses-its-3d-encoder-at-runtime",
    "json_url": "https://themachinepress.com/story/mp-2026-10-15-010.json",
    "first_published_at": "2026-10-15T09:00:00.000-04:00",
    "modified_at": "2026-10-15T09:00:00.000-04:00",
    "content_status": "updated",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-15-010-001",
        "text": "GeoThinker-F learns predictive geometric and future-semantic latents from an offline expert, then removes that expert for inference.",
        "source_ids": [
          "source-2026-10-15-010"
        ],
        "qualification": null
      }
    ],
    "source_ids": [
      "source-2026-10-15-010"
    ],
    "tags": [
      "spatial intelligence",
      "latent prediction",
      "inference efficiency"
    ],
    "image_url": null,
    "corrections": [
      {
        "correction_id": "correction-2026-10-15-001",
        "revision": 2,
        "summary": "Publication note: This October 15 edition was prepared before its scheduled 9:00 AM America/Detroit reveal, but publication was delayed by a production build failure. The original reporting and image-generation timestamps are preserved.",
        "published_at": "2026-10-15T09:00:00.000-04:00",
        "affected_story_ids": [],
        "affected_claim_ids": []
      }
    ]
  },
  "sources": [
    {
      "source_id": "source-2026-10-15-010",
      "title": "arXiv preprint 2610.20592, original v1",
      "publisher": "arXiv / submitting research authors",
      "url": "https://arxiv.org/abs/2610.20592v1",
      "canonical_url": "https://arxiv.org/abs/2610.20592v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-14T13:59:03.000-04:00",
      "accessed_at": "2026-10-15T08:28:29.903-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-15-010-001"
      ]
    }
  ],
  "corrections": [
    {
      "correction_id": "correction-2026-10-15-001",
      "revision": 2,
      "summary": "Publication note: This October 15 edition was prepared before its scheduled 9:00 AM America/Detroit reveal, but publication was delayed by a production build failure. The original reporting and image-generation timestamps are preserved.",
      "published_at": "2026-10-15T09:00:00.000-04:00",
      "affected_story_ids": [],
      "affected_claim_ids": []
    }
  ],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Spatial Model Dismisses Its 3D Encoder at Runtime",
    "publisher": "The Machine Press",
    "published_at": "2026-10-15T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-15-010/the-spatial-model-dismisses-its-3d-encoder-at-runtime"
  }
}