{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-29-008",
    "source_story_id": "tmp-story-tokencast-agent-consumption",
    "edition_id": "mp-2026-09-29-morning-0082",
    "edition_url": "https://themachinepress.com/edition/2026-09-29",
    "position": 8,
    "story_type": "dispatch",
    "section": "infrastructure",
    "editorial_classification": "editorial",
    "headline": "An Agent's Token Budget Was Forecast While It Ran",
    "slug": "an-agent-s-token-budget-was-forecast-while-it-ran",
    "dek": "TokenCast composed segment costs and context growth without making another model call.",
    "summary": "TokenCast composed segment costs and context growth without making another model call.",
    "body_text": "TokenCast records each execution segment's own token use and the context growth it adds, then composes segments to estimate repeated input costs later in a run. On SWE-bench Verified, the authors report a mean cumulative prediction time of 32.8 milliseconds per run. Across four task suites and six agent models, mean absolute error improved by an average 14.5% over the strongest comparator; an offline replay used 21.3% fewer tokens than a fixed-budget policy at matched trace completion. Replay results do not prove identical savings in live production systems.",
    "why_it_matters": "TokenCast composed segment costs and context growth without making another model call.",
    "limitations": [
      "Replay results do not prove identical savings in live production systems."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-29-008/an-agent-s-token-budget-was-forecast-while-it-ran",
    "json_url": "https://themachinepress.com/story/mp-2026-09-29-008.json",
    "first_published_at": "2026-09-29T09:00:00.000-04:00",
    "modified_at": "2026-09-29T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-29-008-001",
        "text": "TokenCast composed segment costs and context growth without making another model call.",
        "source_ids": [
          "source-2026-09-29-008"
        ],
        "qualification": "Replay results do not prove identical savings in live production systems."
      }
    ],
    "source_ids": [
      "source-2026-09-29-008"
    ],
    "tags": [
      "agent infrastructure",
      "token budgets",
      "forecasting"
    ],
    "image_url": "https://themachinepress.com/issues/2026-09-29/tokencast-server-aisle.webp",
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-29-008",
      "title": "arXiv preprint 2609.35760",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.35760",
      "canonical_url": "https://arxiv.org/abs/2609.35760",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-28T13:59:09.000-04:00",
      "accessed_at": "2026-09-29T08:18:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-29-008-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "An Agent's Token Budget Was Forecast While It Ran",
    "publisher": "The Machine Press",
    "published_at": "2026-09-29T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-29-008/an-agent-s-token-budget-was-forecast-while-it-ran"
  }
}