{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-14-006",
    "source_story_id": "tmp-story-tev-dense-agent-rewards",
    "edition_id": "mp-2026-10-14-morning-0096",
    "edition_url": "https://themachinepress.com/edition/2026-10-14",
    "position": 6,
    "story_type": "dispatch",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "Long Tasks Gain a Progress Meter",
    "slug": "long-tasks-gain-a-progress-meter",
    "dek": "Tev-Gym supplies 1,170 verified environments with dense stepwise rewards for training long-horizon agents.",
    "summary": "Tev-Gym supplies 1,170 verified environments with dense stepwise rewards for training long-horizon agents.",
    "body_text": "Long-horizon agent tasks can supply a final pass or fail signal while leaving thousands of intermediate choices unexplained. Tev-Gym begins with a hidden specification of the domain, workflow and expected state, then renders 1,170 environments across seven domains. That latent specification also yields a progress measure, which the authors turn into per-step reward shaping. They report gains for a 27-billion-parameter model on computing, math and terminal benchmarks. Comparisons to named frontier models are claims from the paper's evaluation and depend on its task subsets and Pass@5 protocol. The durable contribution is the training structure: verified progress can make sparse, multi-hour work more learnable.",
    "why_it_matters": "Tev-Gym supplies 1,170 verified environments with dense stepwise rewards for training long-horizon agents.",
    "limitations": [],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-14-006/long-tasks-gain-a-progress-meter",
    "json_url": "https://themachinepress.com/story/mp-2026-10-14-006.json",
    "first_published_at": "2026-10-14T09:00:00.000-04:00",
    "modified_at": "2026-10-14T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-14-006-001",
        "text": "Tev-Gym supplies 1,170 verified environments with dense stepwise rewards for training long-horizon agents.",
        "source_ids": [
          "source-2026-10-14-006"
        ],
        "qualification": null
      }
    ],
    "source_ids": [
      "source-2026-10-14-006"
    ],
    "tags": [
      "Tev-Gym",
      "long-horizon agents",
      "reward shaping"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-14-006",
      "title": "arXiv preprint 2610.18868, original v1",
      "publisher": "arXiv / submitting research authors",
      "url": "https://arxiv.org/abs/2610.18868v1",
      "canonical_url": "https://arxiv.org/abs/2610.18868v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-13T13:59:51.000-04:00",
      "accessed_at": "2026-10-14T08:24:30.364-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-14-006-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Long Tasks Gain a Progress Meter",
    "publisher": "The Machine Press",
    "published_at": "2026-10-14T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-14-006/long-tasks-gain-a-progress-meter"
  }
}