{
  "schema_version": "1.0",
  "publisher": "AI Release",
  "article": {
    "title": "GPT-6.1-Sol and GPT-6-Astra Are Obsessed With Appendices: AI Reviewers Are Ruining Peer Review",
    "canonical_url": "https://ai-release.org/posts/gpt-6-1-sol-i-gpt-6-astra-zalipli-na-prilozheniyah-ii-revyuery-portyat-6e61cb65_en.html",
    "language": "en",
    "article_type": "news",
    "published_at": "2026-11-21T00:00:00+00:00",
    "updated_at": "2026-11-21T00:00:00+00:00"
  },
  "summary": "OpenAI's new models — GPT-6.1-Sol and GPT-6-Astra — pull 50–70% of their review comments from paper appendices rather than the main text. For humans, appendices account for 8.4% of complaints; for AI, it's 62% — a gap of more than seven times.",
  "topics": [
    "OpenAI",
    "Pir-revyu",
    "GPT-6",
    "Nauchnye",
    "Benchmarks",
    "AI",
    "LLM",
    "neural networks"
  ],
  "entities": [
    {
      "id": "E-SRC",
      "name": "www.nytimes.com",
      "type": "source",
      "description": "source domain",
      "url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html"
    },
    {
      "id": "E-T1",
      "name": "OpenAI",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-T2",
      "name": "Pir-revyu",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-T3",
      "name": "GPT-6",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-T4",
      "name": "Nauchnye",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-T5",
      "name": "Benchmarks",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-T6",
      "name": "AI",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-T7",
      "name": "LLM",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-T8",
      "name": "neural networks",
      "type": "topic",
      "description": "publication tag",
      "url": null
    },
    {
      "id": "E-AIR",
      "name": "AI Release",
      "type": "publisher",
      "description": "published the article",
      "url": "https://ai-release.org/"
    }
  ],
  "claims": [
    {
      "id": "AR-GPT61SOLIG-001",
      "statement": "OpenAI's new models — GPT-6.1-Sol and GPT-6-Astra — pull 50–70% of their review comments from paper appendices rather than the main text.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-001",
      "evidence_excerpt": "OpenAI's new models — GPT-6.1-Sol and GPT-6-Astra — pull 50–70% of their review comments from paper appendices rather than the main text."
    },
    {
      "id": "AR-GPT61SOLIG-002",
      "statement": "For humans, appendices account for 8.4% of complaints; for AI, it's 62% — a gap of more than seven times.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-002",
      "evidence_excerpt": "For humans, appendices account for 8.4% of complaints; for AI, it's 62% — a gap of more than seven times."
    },
    {
      "id": "AR-GPT61SOLIG-003",
      "statement": "Side effect: the share of appendix nitpicks in live human reviews has jumped from under 11% to 27.6%.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-003",
      "evidence_excerpt": "Side effect: the share of appendix nitpicks in live human reviews has jumped from under 11% to 27.6%."
    },
    {
      "id": "AR-GPT61SOLIG-004",
      "statement": "A team of researchers ran 2,000 papers from four popular conferences through OpenAI's new models and compared their comments with what human reviewers actually write.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-004",
      "evidence_excerpt": "A team of researchers ran 2,000 papers from four popular conferences through OpenAI's new models and compared their comments with what human reviewers actually write."
    },
    {
      "id": "AR-GPT61SOLIG-005",
      "statement": "The gap turned out to be huge.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-005",
      "evidence_excerpt": "The gap turned out to be huge."
    },
    {
      "id": "AR-GPT61SOLIG-006",
      "statement": "For GPT-5, less than 30% of the criticism targeted appendices; for GPT-6.1-Sol and GPT-6-Astra, it's roughly 50–70% — and the effect holds across all four conferences, so it's not venue-specific.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-006",
      "evidence_excerpt": "For GPT-5, less than 30% of the criticism targeted appendices; for GPT-6.1-Sol and GPT-6-Astra, it's roughly 50–70% — and the effect holds across all four conferences, so it's not venue-specific."
    },
    {
      "id": "AR-GPT61SOLIG-007",
      "statement": "The paper's authors note: \"The difference between generations is large.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-007",
      "evidence_excerpt": "The paper's authors note: \"The difference between generations is large."
    },
    {
      "id": "AR-GPT61SOLIG-008",
      "statement": "For GPT-5, only a small share of weaknesses cited relate to the appendix.",
      "claim_type": "reported_fact",
      "confidence": "not_assessed",
      "source_type": "publication source",
      "verification_status": "reported_unverified",
      "source_url": "https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "article_anchor": "#ai-claim-008",
      "evidence_excerpt": "For GPT-5, only a small share of weaknesses cited relate to the appendix."
    }
  ],
  "relationships": [
    {
      "subject": "www.nytimes.com",
      "predicate": "reports on",
      "object": "GPT-6.1-Sol and GPT-6-Astra Are Obsessed With Appendices: AI Reviewers Are Ruining Peer Review",
      "claim_id": "AR-GPT61SOLIG-001"
    },
    {
      "subject": "AI Release",
      "predicate": "publishes",
      "object": "GPT-6.1-Sol and GPT-6-Astra Are Obsessed With Appendices: AI Reviewers Are Ruining Peer Review",
      "claim_id": "AR-GPT61SOLIG-001"
    }
  ],
  "questions_and_answers": [
    {
      "question": "What is this article about?",
      "answer": "GPT-6.1-Sol and GPT-6-Astra Are Obsessed With Appendices: AI Reviewers Are Ruining Peer Review. OpenAI's new models — GPT-6.1-Sol and GPT-6-Astra — pull 50–70% of their review comments from paper appendices rather than the main text.",
      "claim_ids": [
        "AR-GPT61SOLIG-001"
      ],
      "article_anchor": "#ai-knowledge"
    },
    {
      "question": "What happened?",
      "answer": "OpenAI's new models — GPT-6.1-Sol and GPT-6-Astra — pull 50–70% of their review comments from paper appendices rather than the main text. For humans, appendices account for 8.4% of complaints; for AI, it's 62% — a gap of more than seven times. Side effect: the share of appendix nitpicks in live human reviews has jumped from under 11% to 27.6%.",
      "claim_ids": [
        "AR-GPT61SOLIG-001",
        "AR-GPT61SOLIG-002",
        "AR-GPT61SOLIG-003"
      ],
      "article_anchor": "#ai-knowledge"
    },
    {
      "question": "Who is the source?",
      "answer": "www.nytimes.com: https://www.nytimes.com/2026/11/21/science/outages-tech-history.html",
      "claim_ids": [],
      "article_anchor": "#ai-knowledge"
    },
    {
      "question": "What remains unknown?",
      "answer": "The article reports the cited source; independent verification and full details are not provided.",
      "claim_ids": [],
      "article_anchor": "#ai-knowledge"
    }
  ],
  "limitations": [
    "No independent verification is provided.",
    "Full details, dates and official documents may be missing.",
    "The article reports a source rather than being the primary source."
  ],
  "citation": {
    "preferred_citation": "AI Release. «GPT-6.1-Sol and GPT-6-Astra Are Obsessed With Appendices: AI Reviewers Are Ruining Peer Review». 2026-11-21. https://ai-release.org/posts/gpt-6-1-sol-i-gpt-6-astra-zalipli-na-prilozheniyah-ii-revyuery-portyat-6e61cb65_en.html",
    "canonical_url": "https://ai-release.org/posts/gpt-6-1-sol-i-gpt-6-astra-zalipli-na-prilozheniyah-ii-revyuery-portyat-6e61cb65_en.html",
    "publisher": "AI Release"
  }
}