{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-01-026",
    "source_story_id": "tmp-story-turing-physical-ai-model",
    "edition_id": "mp-2026-09-01-morning-0054",
    "edition_url": "https://themachinepress.com/edition/2026-09-01",
    "position": 15,
    "story_type": "dispatch",
    "section": "frontier-models",
    "editorial_classification": "editorial",
    "headline": "Two Billion Active Parameters Carried a Twenty-Billion Model",
    "slug": "two-billion-active-parameters-carried-a-twenty-billion-model",
    "dek": "Turing-20B-A2B combines dynamic expert routing and hybrid attention for long-context, latency-sensitive physical-AI workloads.",
    "summary": "Turing-20B-A2B combines dynamic expert routing and hybrid attention for long-context, latency-sensitive physical-AI workloads.",
    "body_text": "The mixture-of-experts model activates about two billion of its twenty billion parameters per token, using quantile routing to vary expert allocation while controlling average compute. Lightning Attention is mixed with a small number of full-attention layers; continued pretraining extends native context to 128K and YaRN is used for 512K inference-time extension. The authors report base-model capability above Qwen3-8B Base and near Qwen3.5-9B Base with favorable prefill scaling. Those are self-reported benchmark comparisons, not independent deployment results.",
    "why_it_matters": "Turing-20B-A2B combines dynamic expert routing and hybrid attention for long-context, latency-sensitive physical-AI workloads.",
    "limitations": [
      "Those are self-reported benchmark comparisons, not independent deployment results."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-01-026/two-billion-active-parameters-carried-a-twenty-billion-model",
    "json_url": "https://themachinepress.com/story/mp-2026-09-01-026.json",
    "first_published_at": "2026-09-01T09:00:00.000-04:00",
    "modified_at": "2026-09-01T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-01-026-001",
        "text": "Turing-20B-A2B combines dynamic expert routing and hybrid attention for long-context, latency-sensitive physical-AI workloads.",
        "source_ids": [
          "source-2026-09-01-015"
        ],
        "qualification": "Those are self-reported benchmark comparisons, not independent deployment results."
      }
    ],
    "source_ids": [
      "source-2026-09-01-015"
    ],
    "tags": [
      "mixture of experts",
      "long context",
      "physical AI"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-01-015",
      "title": "arXiv preprint 2608.30567",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2608.30567",
      "canonical_url": "https://arxiv.org/abs/2608.30567",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-08-31T06:42:35.000-04:00",
      "accessed_at": "2026-09-01T08:25:35.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-01-026-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Two Billion Active Parameters Carried a Twenty-Billion Model",
    "publisher": "The Machine Press",
    "published_at": "2026-09-01T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-01-026/two-billion-active-parameters-carried-a-twenty-billion-model"
  }
}