{
  "datasetId": "ai-search-citation-benchmark-snapshot-2026-07-22",
  "schemaVersion": "1.1.0",
  "createdAt": "2026-07-22",
  "language": "zh-CN",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "licenseScope": "DragonAI's observation design, run-level citation URL records, aggregation, field definitions, and Chinese descriptions. Complete model answers, provider API payloads, third-party page content, product names, and trademarks are excluded from the license grant.",
  "disclosureZh": "DragonAI 运营 AI 产品经理课程并监测自有域名。本数据集保留目标引用和品牌提及均为零的结果，不把发布、抓取、提交或等待时长计为 GEO 成功。",
  "providerRelationshipBoundaryZh": "千问、Kimi、Gemini 及模型名称属于各自权利人的产品标识；本数据集不暗示 DragonAI 与任何提供商存在合作、认可或背书关系。",
  "thirdPartyRightsBoundaryZh": "引用 URL 仅作为点时观察记录，不构成对第三方内容的复制、许可、准确性确认或推荐，也不授予任何第三方页面内容、产品名称或商标的权利。",
  "researchDesign": {
    "explorationSampleA": {
      "purposeZh": "观察技术选型问题的跨平台来源分布；不进入正式趋势。",
      "query": "RAG和模型微调应该怎么选择？请给出决策条件、验证实验和可靠来源。",
      "targetPath": "/comparisons/rag-vs-fine-tuning",
      "publicDatasetUrl": "https://course.dragonai.tech/datasets/ai-search-rag-finetuning-citation-sample-2026-07.json",
      "completedRuns": 9,
      "comparableToWindows": []
    },
    "formalBenchmarkBC": {
      "purposeZh": "用固定 GEO 元问题观察研究页的无品牌引用可见性。",
      "queryId": "benchmark-ai-search-rag-finetuning-citations",
      "query": "千问、Kimi、Gemini回答RAG和微调选型问题时通常引用哪些来源？请给出可复核样本、平台差异和研究限制。",
      "targetDomain": "course.dragonai.tech",
      "targetPaths": [
        "/research/ai-search-rag-finetuning-citations-2026-07"
      ],
      "baselineWindowId": "B",
      "postpublicationWindowId": "C",
      "comparisonBoundaryZh": "只有 B 与 C 使用同一查询；模型相同的 Qwen 与 Gemini 可进入严格趋势，Kimi 因模型迁移只建立 C 窗口新基线。A 不与 B 或 C比较引用数量或来源分布。"
    }
  },
  "benchmarkGate": {
    "benchmarkReady": true,
    "readinessReason": "elapsed_release_window_without_verified_target_crawl",
    "postCrawlBenchmarkReady": false,
    "timeboxedBenchmarkReady": true,
    "interpretationZh": "24 小时时间盒仅授权不可归因的点时复测；它不是目标页抓取、索引、引用或排名证据，也不能证明页面发布或改写造成结果变化。"
  },
  "methodology": {
    "providers": [
      "gemini_google_search_grounding",
      "kimi_builtin_web_search",
      "qwen_dashscope_web_search"
    ],
    "repetitionsPerProviderPerWindow": 3,
    "samePromptAcrossFormalWindows": true,
    "nativeWebSearchRequired": true,
    "concreteUrlRuleZh": "只保留含 http/https、可解析主机名且不含星号、省略号或截断说明的具体 URL；被丢弃的展示占位符只计数，不作为引用或竞争域证据。",
    "targetAttributionRuleZh": "只有具体 URL 的主机名等于 course.dragonai.tech 或其子域名才计为目标引用。",
    "brandMentionRuleZh": "回答正文出现预先登记的 DragonAI、烛龙智元或目标域名实体时计为品牌提及；品牌提及与目标 URL 引用分开记录。",
    "rawAnswerPublication": "excluded",
    "providerApiPayloadPublication": "excluded"
  },
  "windows": [
    {
      "windowId": "B",
      "phase": "prepublication",
      "reportGeneratedAt": "2026-07-21T00:04:10+00:00",
      "observationBatches": {
        "gemini_google_search_grounding": "2026-07-21T00:02:51+00:00",
        "kimi_builtin_web_search": "2026-07-21T00:02:07+00:00",
        "qwen_dashscope_web_search": "2026-07-21T00:00:46+00:00"
      },
      "aggregate": {
        "completedRuns": 9,
        "verifiableRuns": 9,
        "concreteUrlOccurrences": 106,
        "discardedPlaceholderCount": 0,
        "distinctConcreteUrls": 60,
        "distinctHostnames": 39,
        "targetCitationRuns": 0,
        "brandMentionRuns": 0,
        "hostnameSharedByAllThreeProviders": [],
        "countsAsGeoSuccess": false
      },
      "providerResults": [
        {
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "completedRuns": 3,
          "verifiableCitationSurface": true,
          "attributableCitationRuns": 3,
          "citationCountsByRun": [
            24,
            21,
            23
          ],
          "discardedPlaceholderCount": 0,
          "concreteUrlOccurrences": 68,
          "distinctConcreteUrls": 40,
          "distinctHostnames": 29,
          "targetCitationRuns": 0,
          "brandMentionRuns": 0,
          "hostnameOccurrences": [
            {
              "hostname": "medium.com",
              "citationOccurrences": 7
            },
            {
              "hostname": "cloud.google.com",
              "citationOccurrences": 5
            },
            {
              "hostname": "docs.cloud.google.com",
              "citationOccurrences": 4
            },
            {
              "hostname": "arxiv.org",
              "citationOccurrences": 3
            },
            {
              "hostname": "feba-systeme.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "memgraph.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "winder.ai",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.alibabacloud.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.databricks.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.ibm.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.infoworld.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.layer3labs.io",
              "citationOccurrences": 3
            },
            {
              "hostname": "learnaitogethernewsletter.substack.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.reddit.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.redhat.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "cohere.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "kartaca.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.matillion.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.mindstudio.ai",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.scifiniti.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "machinelearningmastery.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "neo4j.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "pmc.ncbi.nlm.nih.gov",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.k2view.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.thenile.dev",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.actian.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.techtarget.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "platform.kimi.ai",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.authorea.com",
              "citationOccurrences": 1
            }
          ],
          "repeatedUrls": [
            {
              "url": "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
              "runOccurrences": 3
            },
            {
              "url": "https://feba-systeme.com/the-real-limits-of-rag-when-you-need-fine-tuning-instead",
              "runOccurrences": 3
            },
            {
              "url": "https://medium.com/@iamshobhitagarwal/kimi-k3-the-complete-engineers-guide-benchmarks-pricing-code-and-when-to-actually-use-it-c0faf61eaa0a",
              "runOccurrences": 3
            },
            {
              "url": "https://memgraph.com/blog/llm-limitations-fine-tuning-vs-rag",
              "runOccurrences": 3
            },
            {
              "url": "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
              "runOccurrences": 3
            },
            {
              "url": "https://www.databricks.com/blog/rag-vs-fine-tuning",
              "runOccurrences": 3
            },
            {
              "url": "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
              "runOccurrences": 3
            },
            {
              "url": "https://www.infoworld.com/article/2336988/the-limitations-of-model-fine-tuning-and-rag.html",
              "runOccurrences": 3
            },
            {
              "url": "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
              "runOccurrences": 3
            },
            {
              "url": "https://learnaitogethernewsletter.substack.com/p/lai-76-qwen-fine-tuning-real-time",
              "runOccurrences": 2
            },
            {
              "url": "https://www.alibabacloud.com/blog/bridging-the-gap-in-generative-ai-fine-tuning-meets-rag-on-alibaba-cloud_602142",
              "runOccurrences": 2
            },
            {
              "url": "https://www.reddit.com/r/LocalLLaMA/comments/1ouxy9y/finetune_for_rag/",
              "runOccurrences": 2
            },
            {
              "url": "https://www.redhat.com/en/topics/ai/rag-vs-fine-tuning",
              "runOccurrences": 2
            },
            {
              "url": "https://cloud.google.com/blog/products/ai-machine-learning/supervised-fine-tuning-for-gemini-llm",
              "runOccurrences": 2
            },
            {
              "url": "https://cohere.com/blog/rag-vs-fine-tuning",
              "runOccurrences": 2
            },
            {
              "url": "https://kartaca.com/en/fine-tuning-gemini-models-unlocking-industry-specific-intelligence/",
              "runOccurrences": 2
            },
            {
              "url": "https://www.matillion.com/blog/rag-vs-fine-tuning-enterprise-ai-strategy-guide",
              "runOccurrences": 2
            },
            {
              "url": "https://www.mindstudio.ai/blog/open-weight-ai-frontier-kimi-k3-agent-stack",
              "runOccurrences": 2
            },
            {
              "url": "https://www.scifiniti.com/3104-4719/3/2026.0031",
              "runOccurrences": 2
            }
          ]
        },
        {
          "provider": "kimi_builtin_web_search",
          "model": "moonshot-v1-128k",
          "completedRuns": 3,
          "verifiableCitationSurface": true,
          "attributableCitationRuns": 3,
          "citationCountsByRun": [
            0,
            6,
            5
          ],
          "discardedPlaceholderCount": 0,
          "concreteUrlOccurrences": 11,
          "distinctConcreteUrls": 11,
          "distinctHostnames": 4,
          "targetCitationRuns": 0,
          "brandMentionRuns": 0,
          "hostnameOccurrences": [
            {
              "hostname": "arxiv.org",
              "citationOccurrences": 6
            },
            {
              "hostname": "mp.weixin.qq.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "beta.openai.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.hubwiz.com",
              "citationOccurrences": 1
            }
          ],
          "repeatedUrls": []
        },
        {
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "completedRuns": 3,
          "verifiableCitationSurface": true,
          "attributableCitationRuns": 3,
          "citationCountsByRun": [
            9,
            9,
            9
          ],
          "discardedPlaceholderCount": 0,
          "concreteUrlOccurrences": 27,
          "distinctConcreteUrls": 9,
          "distinctHostnames": 7,
          "targetCitationRuns": 0,
          "brandMentionRuns": 0,
          "hostnameOccurrences": [
            {
              "hostname": "zhuanlan.zhihu.com",
              "citationOccurrences": 9
            },
            {
              "hostname": "cloud.tencent.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "developer.cloud.tencent.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "hub.baai.ac.cn",
              "citationOccurrences": 3
            },
            {
              "hostname": "tech.ifeng.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.163.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.thepaper.cn",
              "citationOccurrences": 3
            }
          ],
          "repeatedUrls": [
            {
              "url": "https://cloud.tencent.com/developer/article/2589138",
              "runOccurrences": 3
            },
            {
              "url": "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
              "runOccurrences": 3
            },
            {
              "url": "https://hub.baai.ac.cn/view/35723",
              "runOccurrences": 3
            },
            {
              "url": "https://tech.ifeng.com/c/8uuO6XHofwJ",
              "runOccurrences": 3
            },
            {
              "url": "https://www.163.com/dy/article/KO7GBD17051189P5.html",
              "runOccurrences": 3
            },
            {
              "url": "https://www.thepaper.cn/newsDetail_forward_32034602",
              "runOccurrences": 3
            },
            {
              "url": "https://zhuanlan.zhihu.com/p/2035991209609576779",
              "runOccurrences": 3
            },
            {
              "url": "https://zhuanlan.zhihu.com/p/2060028263477540134",
              "runOccurrences": 3
            },
            {
              "url": "https://zhuanlan.zhihu.com/p/2061520970595275127",
              "runOccurrences": 3
            }
          ]
        }
      ],
      "runRecords": [
        {
          "runIndex": 1,
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "observedAt": "2026-07-21T00:02:51+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 24,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://arxiv.org/abs/2510.01375",
            "https://arxiv.org/html/2506.22852v1",
            "https://cloud.google.com/transform/top-five-gen-ai-tuning-use-cases-gemini-hundreds-of-orgs",
            "https://cloud.google.com/use-cases/fine-tuning-ai-models",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
            "https://feba-systeme.com/the-real-limits-of-rag-when-you-need-fine-tuning-instead",
            "https://learnaitogethernewsletter.substack.com/p/lai-76-qwen-fine-tuning-real-time",
            "https://machinelearningmastery.com/understanding-rag-part-ix-fine-tuning-llms-for-rag/",
            "https://medium.com/@iamshobhitagarwal/kimi-k3-the-complete-engineers-guide-benchmarks-pricing-code-and-when-to-actually-use-it-c0faf61eaa0a",
            "https://medium.com/@vrajdcs/rag-and-fine-tuning-implementation-best-practices-for-scalable-and-accurate-genai-solutions-e038ea957512",
            "https://medium.com/google-cloud/fine-tuning-gemini-best-practices-for-data-hyperparameters-and-evaluation-65f7c7b6b15f",
            "https://memgraph.com/blog/llm-limitations-fine-tuning-vs-rag",
            "https://neo4j.com/blog/developer/fine-tuning-vs-rag/",
            "https://pmc.ncbi.nlm.nih.gov/articles/PMC12938813/",
            "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
            "https://www.alibabacloud.com/blog/bridging-the-gap-in-generative-ai-fine-tuning-meets-rag-on-alibaba-cloud_602142",
            "https://www.databricks.com/blog/rag-vs-fine-tuning",
            "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
            "https://www.infoworld.com/article/2336988/the-limitations-of-model-fine-tuning-and-rag.html",
            "https://www.k2view.com/blog/retrieval-augmented-generation-vs-fine-tuning/",
            "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
            "https://www.reddit.com/r/LocalLLaMA/comments/1ouxy9y/finetune_for_rag/",
            "https://www.redhat.com/en/topics/ai/rag-vs-fine-tuning",
            "https://www.thenile.dev/blog/fine-tuning-or-rag"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 2,
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "observedAt": "2026-07-21T00:02:51+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 21,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://arxiv.org/abs/2401.08406",
            "https://cloud.google.com/blog/products/ai-machine-learning/supervised-fine-tuning-for-gemini-llm",
            "https://cloud.google.com/blog/products/ai-machine-learning/to-tune-or-not-to-tune-a-guide-to-leveraging-your-data-with-llms",
            "https://cohere.com/blog/rag-vs-fine-tuning",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
            "https://feba-systeme.com/the-real-limits-of-rag-when-you-need-fine-tuning-instead",
            "https://kartaca.com/en/fine-tuning-gemini-models-unlocking-industry-specific-intelligence/",
            "https://medium.com/@iamshobhitagarwal/kimi-k3-the-complete-engineers-guide-benchmarks-pricing-code-and-when-to-actually-use-it-c0faf61eaa0a",
            "https://memgraph.com/blog/llm-limitations-fine-tuning-vs-rag",
            "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
            "https://www.actian.com/blog/databases/should-you-use-rag-or-fine-tune-your-llm/",
            "https://www.alibabacloud.com/help/en/model-studio/text-generation-model-tuning",
            "https://www.databricks.com/blog/rag-vs-fine-tuning",
            "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
            "https://www.infoworld.com/article/2336988/the-limitations-of-model-fine-tuning-and-rag.html",
            "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
            "https://www.matillion.com/blog/rag-vs-fine-tuning-enterprise-ai-strategy-guide",
            "https://www.mindstudio.ai/blog/open-weight-ai-frontier-kimi-k3-agent-stack",
            "https://www.redhat.com/en/topics/ai/rag-vs-fine-tuning",
            "https://www.scifiniti.com/3104-4719/3/2026.0031",
            "https://www.techtarget.com/searchenterpriseai/tip/Understanding-the-limitations-and-challenges-of-RAG-systems"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 3,
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "observedAt": "2026-07-21T00:02:51+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 23,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://cloud.google.com/blog/products/ai-machine-learning/supervised-fine-tuning-for-gemini-llm",
            "https://cohere.com/blog/rag-vs-fine-tuning",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning/use",
            "https://feba-systeme.com/the-real-limits-of-rag-when-you-need-fine-tuning-instead",
            "https://kartaca.com/en/fine-tuning-gemini-models-unlocking-industry-specific-intelligence/",
            "https://learnaitogethernewsletter.substack.com/p/lai-76-qwen-fine-tuning-real-time",
            "https://medium.com/@candemir13/fine-tuning-vs-rag-a-decision-framework-for-practitioners-7c26cba89768",
            "https://medium.com/@iamshobhitagarwal/kimi-k3-the-complete-engineers-guide-benchmarks-pricing-code-and-when-to-actually-use-it-c0faf61eaa0a",
            "https://medium.com/google-cloud/a-step-by-step-guide-to-fine-tuning-gemini-for-question-answering-8b3fb117dbbf",
            "https://memgraph.com/blog/llm-limitations-fine-tuning-vs-rag",
            "https://platform.kimi.ai/docs/guide/kimi-k3-quickstart",
            "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
            "https://www.alibabacloud.com/blog/bridging-the-gap-in-generative-ai-fine-tuning-meets-rag-on-alibaba-cloud_602142",
            "https://www.authorea.com/doi/10.22541/au.176903295.54191070",
            "https://www.databricks.com/blog/rag-vs-fine-tuning",
            "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
            "https://www.infoworld.com/article/2336988/the-limitations-of-model-fine-tuning-and-rag.html",
            "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
            "https://www.matillion.com/blog/rag-vs-fine-tuning-enterprise-ai-strategy-guide",
            "https://www.mindstudio.ai/blog/open-weight-ai-frontier-kimi-k3-agent-stack",
            "https://www.reddit.com/r/LocalLLaMA/comments/1ouxy9y/finetune_for_rag/",
            "https://www.scifiniti.com/3104-4719/3/2026.0031"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 1,
          "provider": "kimi_builtin_web_search",
          "model": "moonshot-v1-128k",
          "observedAt": "2026-07-21T00:02:07+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 0,
          "discardedPlaceholderCount": 0,
          "citedUrls": [],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 2,
          "provider": "kimi_builtin_web_search",
          "model": "moonshot-v1-128k",
          "observedAt": "2026-07-21T00:02:07+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 6,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "http://mp.weixin.qq.com/s?__biz=MzIyMzU4NDMyOQ==&mid=2247507933&idx=8&sn=f907772c1a8be55f3721c27ef78f5669",
            "https://arxiv.org/abs/2005.11401",
            "https://arxiv.org/abs/2106.09685",
            "https://arxiv.org/abs/2403.05530)等",
            "https://arxiv.org/abs/2404.16130",
            "https://beta.openai.com/docs/)作为RAG技术的一个参考来源"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 3,
          "provider": "kimi_builtin_web_search",
          "model": "moonshot-v1-128k",
          "observedAt": "2026-07-21T00:02:07+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 5,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "http://mp.weixin.qq.com/s?__biz=MzIyMDYzMTE4MQ==&mid=2247486102&idx=1&sn=9c4396e0756f51460f531a45da95730d",
            "http://mp.weixin.qq.com/s?__biz=MzU1NTUxNTM0Mg==&mid=2247567307&idx=2&sn=a22d95bd843ec3d2e8d42a3f8f679f10",
            "http://www.hubwiz.com/blog/gemini-fine-tuning-on-squad-dataset/",
            "https://arxiv.org/pdf/2407.08223v2",
            "https://arxiv.org/pdf/2410.05983"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 1,
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "observedAt": "2026-07-21T00:00:46+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 9,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://cloud.tencent.com/developer/article/2589138",
            "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
            "https://hub.baai.ac.cn/view/35723",
            "https://tech.ifeng.com/c/8uuO6XHofwJ",
            "https://www.163.com/dy/article/KO7GBD17051189P5.html",
            "https://www.thepaper.cn/newsDetail_forward_32034602",
            "https://zhuanlan.zhihu.com/p/2035991209609576779",
            "https://zhuanlan.zhihu.com/p/2060028263477540134",
            "https://zhuanlan.zhihu.com/p/2061520970595275127"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 2,
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "observedAt": "2026-07-21T00:00:46+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 9,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://cloud.tencent.com/developer/article/2589138",
            "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
            "https://hub.baai.ac.cn/view/35723",
            "https://tech.ifeng.com/c/8uuO6XHofwJ",
            "https://www.163.com/dy/article/KO7GBD17051189P5.html",
            "https://www.thepaper.cn/newsDetail_forward_32034602",
            "https://zhuanlan.zhihu.com/p/2035991209609576779",
            "https://zhuanlan.zhihu.com/p/2060028263477540134",
            "https://zhuanlan.zhihu.com/p/2061520970595275127"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 3,
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "observedAt": "2026-07-21T00:00:46+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 9,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://cloud.tencent.com/developer/article/2589138",
            "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
            "https://hub.baai.ac.cn/view/35723",
            "https://tech.ifeng.com/c/8uuO6XHofwJ",
            "https://www.163.com/dy/article/KO7GBD17051189P5.html",
            "https://www.thepaper.cn/newsDetail_forward_32034602",
            "https://zhuanlan.zhihu.com/p/2035991209609576779",
            "https://zhuanlan.zhihu.com/p/2060028263477540134",
            "https://zhuanlan.zhihu.com/p/2061520970595275127"
          ],
          "targetCited": false,
          "brandMentioned": false
        }
      ]
    },
    {
      "windowId": "C",
      "phase": "postpublication",
      "reportGeneratedAt": "2026-07-22T01:49:50+00:00",
      "observationBatches": {
        "gemini_google_search_grounding": "2026-07-22T01:31:21+00:00",
        "kimi_builtin_web_search": "2026-07-22T01:34:12+00:00",
        "qwen_dashscope_web_search": "2026-07-22T01:29:56+00:00"
      },
      "aggregate": {
        "completedRuns": 9,
        "verifiableRuns": 9,
        "concreteUrlOccurrences": 124,
        "discardedPlaceholderCount": 2,
        "distinctConcreteUrls": 74,
        "distinctHostnames": 50,
        "targetCitationRuns": 0,
        "brandMentionRuns": 0,
        "hostnameSharedByAllThreeProviders": [],
        "countsAsGeoSuccess": false
      },
      "providerResults": [
        {
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "completedRuns": 3,
          "verifiableCitationSurface": true,
          "attributableCitationRuns": 3,
          "citationCountsByRun": [
            23,
            28,
            28
          ],
          "discardedPlaceholderCount": 0,
          "concreteUrlOccurrences": 79,
          "distinctConcreteUrls": 47,
          "distinctHostnames": 34,
          "targetCitationRuns": 0,
          "brandMentionRuns": 0,
          "hostnameOccurrences": [
            {
              "hostname": "medium.com",
              "citationOccurrences": 10
            },
            {
              "hostname": "docs.cloud.google.com",
              "citationOccurrences": 6
            },
            {
              "hostname": "arxiv.org",
              "citationOccurrences": 4
            },
            {
              "hostname": "www.reddit.com",
              "citationOccurrences": 4
            },
            {
              "hostname": "cloud.google.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "learnaitogethernewsletter.substack.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "winder.ai",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.glean.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.ibm.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.layer3labs.io",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.mindstudio.ai",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.scifiniti.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "cohere.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "dev.to",
              "citationOccurrences": 2
            },
            {
              "hostname": "feba-systeme.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "kairntech.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "memgraph.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.databricks.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.infoworld.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.alibabacloud.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "www.authorea.com",
              "citationOccurrences": 2
            },
            {
              "hostname": "community.ibm.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "towardsdatascience.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "cobusgreyling.medium.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "machinelearningmastery.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "sulbhajain.medium.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.matillion.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "daily.dev",
              "citationOccurrences": 1
            },
            {
              "hostname": "huggingface.co",
              "citationOccurrences": 1
            },
            {
              "hostname": "pmc.ncbi.nlm.nih.gov",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.actian.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.microsoft.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.redhat.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.trickywombat.ai",
              "citationOccurrences": 1
            }
          ],
          "repeatedUrls": [
            {
              "url": "https://cloud.google.com/blog/products/ai-machine-learning/supervised-fine-tuning-for-gemini-llm",
              "runOccurrences": 3
            },
            {
              "url": "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
              "runOccurrences": 3
            },
            {
              "url": "https://learnaitogethernewsletter.substack.com/p/lai-76-qwen-fine-tuning-real-time",
              "runOccurrences": 3
            },
            {
              "url": "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
              "runOccurrences": 3
            },
            {
              "url": "https://www.glean.com/blog/retrieval-augemented-generation-vs-fine-tuning",
              "runOccurrences": 3
            },
            {
              "url": "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
              "runOccurrences": 3
            },
            {
              "url": "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
              "runOccurrences": 3
            },
            {
              "url": "https://www.mindstudio.ai/blog/kimi-k3-vs-claude-fable-5-agentic-coding",
              "runOccurrences": 3
            },
            {
              "url": "https://www.scifiniti.com/3104-4719/3/2026.0031",
              "runOccurrences": 3
            },
            {
              "url": "https://arxiv.org/abs/2401.08406",
              "runOccurrences": 2
            },
            {
              "url": "https://cohere.com/blog/rag-vs-fine-tuning",
              "runOccurrences": 2
            },
            {
              "url": "https://dev.to/synsun/rag-vs-fine-tuning-what-i-actually-learned-after-6-months-of-building-llm-apps-1iac",
              "runOccurrences": 2
            },
            {
              "url": "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning/use",
              "runOccurrences": 2
            },
            {
              "url": "https://feba-systeme.com/the-real-limits-of-rag-when-you-need-fine-tuning-instead",
              "runOccurrences": 2
            },
            {
              "url": "https://kairntech.com/blog/articles/retrieval-augmented-generation-vs-fine-tuning-choosing-the-right-approach/",
              "runOccurrences": 2
            },
            {
              "url": "https://medium.com/google-cloud/a-step-by-step-guide-to-fine-tuning-gemini-for-question-answering-8b3fb117dbbf",
              "runOccurrences": 2
            },
            {
              "url": "https://medium.com/google-cloud/fine-tuning-gemini-best-practices-for-data-hyperparameters-and-evaluation-65f7c7b6b15f",
              "runOccurrences": 2
            },
            {
              "url": "https://memgraph.com/blog/llm-limitations-fine-tuning-vs-rag",
              "runOccurrences": 2
            },
            {
              "url": "https://www.databricks.com/blog/rag-vs-fine-tuning",
              "runOccurrences": 2
            },
            {
              "url": "https://www.infoworld.com/article/2336988/the-limitations-of-model-fine-tuning-and-rag.html",
              "runOccurrences": 2
            },
            {
              "url": "https://www.alibabacloud.com/blog/bridging-the-gap-in-generative-ai-fine-tuning-meets-rag-on-alibaba-cloud_602142",
              "runOccurrences": 2
            },
            {
              "url": "https://www.authorea.com/doi/10.22541/au.176903295.54191070",
              "runOccurrences": 2
            },
            {
              "url": "https://www.reddit.com/r/Rag/comments/1ouz0zs/finetune_for_rag/",
              "runOccurrences": 2
            }
          ]
        },
        {
          "provider": "kimi_builtin_web_search",
          "model": "kimi-k2.6",
          "completedRuns": 3,
          "verifiableCitationSurface": true,
          "attributableCitationRuns": 3,
          "citationCountsByRun": [
            0,
            0,
            18
          ],
          "discardedPlaceholderCount": 2,
          "concreteUrlOccurrences": 18,
          "distinctConcreteUrls": 18,
          "distinctHostnames": 11,
          "targetCitationRuns": 0,
          "brandMentionRuns": 0,
          "hostnameOccurrences": [
            {
              "hostname": "arxiv.org",
              "citationOccurrences": 7
            },
            {
              "hostname": "platform.moonshot.cn",
              "citationOccurrences": 2
            },
            {
              "hostname": "36kr.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "ai.google.dev",
              "citationOccurrences": 1
            },
            {
              "hostname": "blog.google",
              "citationOccurrences": 1
            },
            {
              "hostname": "help.aliyun.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "mp.weixin.qq.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.bilibili.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.jiqizhixin.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "www.mckinsey.com",
              "citationOccurrences": 1
            },
            {
              "hostname": "zhuanlan.zhihu.com",
              "citationOccurrences": 1
            }
          ],
          "repeatedUrls": []
        },
        {
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "completedRuns": 3,
          "verifiableCitationSurface": true,
          "attributableCitationRuns": 3,
          "citationCountsByRun": [
            9,
            9,
            9
          ],
          "discardedPlaceholderCount": 0,
          "concreteUrlOccurrences": 27,
          "distinctConcreteUrls": 9,
          "distinctHostnames": 7,
          "targetCitationRuns": 0,
          "brandMentionRuns": 0,
          "hostnameOccurrences": [
            {
              "hostname": "zhuanlan.zhihu.com",
              "citationOccurrences": 9
            },
            {
              "hostname": "cloud.tencent.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "developer.cloud.tencent.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "hub.baai.ac.cn",
              "citationOccurrences": 3
            },
            {
              "hostname": "tech.ifeng.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.163.com",
              "citationOccurrences": 3
            },
            {
              "hostname": "www.thepaper.cn",
              "citationOccurrences": 3
            }
          ],
          "repeatedUrls": [
            {
              "url": "https://cloud.tencent.com/developer/article/2589138",
              "runOccurrences": 3
            },
            {
              "url": "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
              "runOccurrences": 3
            },
            {
              "url": "https://hub.baai.ac.cn/view/35723",
              "runOccurrences": 3
            },
            {
              "url": "https://tech.ifeng.com/c/8uuO6XHofwJ",
              "runOccurrences": 3
            },
            {
              "url": "https://www.163.com/dy/article/KO7GBD17051189P5.html",
              "runOccurrences": 3
            },
            {
              "url": "https://www.thepaper.cn/newsDetail_forward_32034602",
              "runOccurrences": 3
            },
            {
              "url": "https://zhuanlan.zhihu.com/p/2035991209609576779",
              "runOccurrences": 3
            },
            {
              "url": "https://zhuanlan.zhihu.com/p/2060028263477540134",
              "runOccurrences": 3
            },
            {
              "url": "https://zhuanlan.zhihu.com/p/2061520970595275127",
              "runOccurrences": 3
            }
          ]
        }
      ],
      "runRecords": [
        {
          "runIndex": 1,
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "observedAt": "2026-07-22T01:31:21+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 23,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://arxiv.org/abs/2401.08406",
            "https://cloud.google.com/blog/products/ai-machine-learning/supervised-fine-tuning-for-gemini-llm",
            "https://cohere.com/blog/rag-vs-fine-tuning",
            "https://community.ibm.com/community/user/blogs/wendy-munoz/2025/11/28/rag-vs-fine-tuning-best-practices-using-the-ibm-ai",
            "https://dev.to/synsun/rag-vs-fine-tuning-what-i-actually-learned-after-6-months-of-building-llm-apps-1iac",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning/use",
            "https://feba-systeme.com/the-real-limits-of-rag-when-you-need-fine-tuning-instead",
            "https://kairntech.com/blog/articles/retrieval-augmented-generation-vs-fine-tuning-choosing-the-right-approach/",
            "https://learnaitogethernewsletter.substack.com/p/lai-76-qwen-fine-tuning-real-time",
            "https://medium.com/google-cloud/a-step-by-step-guide-to-fine-tuning-gemini-for-question-answering-8b3fb117dbbf",
            "https://medium.com/google-cloud/fine-tuning-gemini-best-practices-for-data-hyperparameters-and-evaluation-65f7c7b6b15f",
            "https://medium.com/leboncoin-tech-blog/how-1-hour-of-fine-tuning-beat-3-weeks-of-rag-engineering-084dbecee49c",
            "https://memgraph.com/blog/llm-limitations-fine-tuning-vs-rag",
            "https://towardsdatascience.com/how-to-make-your-llm-more-accurate-with-rag-fine-tuning/",
            "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
            "https://www.databricks.com/blog/rag-vs-fine-tuning",
            "https://www.glean.com/blog/retrieval-augemented-generation-vs-fine-tuning",
            "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
            "https://www.infoworld.com/article/2336988/the-limitations-of-model-fine-tuning-and-rag.html",
            "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
            "https://www.mindstudio.ai/blog/kimi-k3-vs-claude-fable-5-agentic-coding",
            "https://www.scifiniti.com/3104-4719/3/2026.0031"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 2,
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "observedAt": "2026-07-22T01:31:21+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 28,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://arxiv.org/abs/2401.08406",
            "https://cloud.google.com/blog/products/ai-machine-learning/supervised-fine-tuning-for-gemini-llm",
            "https://cobusgreyling.medium.com/rag-fine-tuning-e541512e9601",
            "https://cohere.com/blog/rag-vs-fine-tuning",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning/use",
            "https://feba-systeme.com/the-real-limits-of-rag-when-you-need-fine-tuning-instead",
            "https://kairntech.com/blog/articles/retrieval-augmented-generation-vs-fine-tuning-choosing-the-right-approach/",
            "https://learnaitogethernewsletter.substack.com/p/lai-76-qwen-fine-tuning-real-time",
            "https://machinelearningmastery.com/understanding-rag-part-ix-fine-tuning-llms-for-rag/",
            "https://medium.com/@nay1228/rag-integration-and-fine-tuning-a-comprehensive-guide-df83894ebeca",
            "https://medium.com/@pranjalkhadka/paper-breakdown-fine-tuning-vs-rag-for-less-popular-knowledge-ef018d2e275c",
            "https://medium.com/@vrajdcs/rag-and-fine-tuning-implementation-best-practices-for-scalable-and-accurate-genai-solutions-e038ea957512",
            "https://medium.com/google-cloud/a-step-by-step-guide-to-fine-tuning-gemini-for-question-answering-8b3fb117dbbf",
            "https://memgraph.com/blog/llm-limitations-fine-tuning-vs-rag",
            "https://sulbhajain.medium.com/fine-tuning-a-custom-model-with-gemini-f94809086bf6",
            "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
            "https://www.alibabacloud.com/blog/bridging-the-gap-in-generative-ai-fine-tuning-meets-rag-on-alibaba-cloud_602142",
            "https://www.authorea.com/doi/10.22541/au.176903295.54191070",
            "https://www.databricks.com/blog/rag-vs-fine-tuning",
            "https://www.glean.com/blog/retrieval-augemented-generation-vs-fine-tuning",
            "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
            "https://www.infoworld.com/article/2336988/the-limitations-of-model-fine-tuning-and-rag.html",
            "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
            "https://www.matillion.com/blog/rag-vs-fine-tuning-enterprise-ai-strategy-guide",
            "https://www.mindstudio.ai/blog/kimi-k3-vs-claude-fable-5-agentic-coding",
            "https://www.reddit.com/r/Rag/comments/1ouz0zs/finetune_for_rag/",
            "https://www.scifiniti.com/3104-4719/3/2026.0031"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 3,
          "provider": "gemini_google_search_grounding",
          "model": "gemini-2.5-flash",
          "observedAt": "2026-07-22T01:31:21+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 28,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://arxiv.org/abs/2510.01600",
            "https://arxiv.org/html/2412.08179",
            "https://cloud.google.com/blog/products/ai-machine-learning/supervised-fine-tuning-for-gemini-llm",
            "https://daily.dev/posts/rag-vs-fine-tuning-when-to-use-which-and-why-it-actually-matters-bnlltukcc",
            "https://dev.to/synsun/rag-vs-fine-tuning-what-i-actually-learned-after-6-months-of-building-llm-apps-1iac",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/rag-engine/fine-tune-rag-transformations",
            "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning",
            "https://huggingface.co/blog/airabbitX/rag-vs-fine-tuning-for-llms-a-com",
            "https://learnaitogethernewsletter.substack.com/p/lai-76-qwen-fine-tuning-real-time",
            "https://medium.com/@airabbitX/20-real-world-examples-of-using-rag-vs-fine-tuning-e24056000251",
            "https://medium.com/google-cloud/building-a-rag-pipeline-with-gemini-2-5-91546c2ceaf6",
            "https://medium.com/google-cloud/fine-tuning-gemini-best-practices-for-data-hyperparameters-and-evaluation-65f7c7b6b15f",
            "https://pmc.ncbi.nlm.nih.gov/articles/PMC12292519/",
            "https://winder.ai/rag-vs-fine-tuning-2026-decision-framework/",
            "https://www.actian.com/blog/databases/should-you-use-rag-or-fine-tune-your-llm/",
            "https://www.alibabacloud.com/blog/bridging-the-gap-in-generative-ai-fine-tuning-meets-rag-on-alibaba-cloud_602142",
            "https://www.authorea.com/doi/10.22541/au.176903295.54191070",
            "https://www.glean.com/blog/retrieval-augemented-generation-vs-fine-tuning",
            "https://www.ibm.com/think/topics/rag-vs-fine-tuning",
            "https://www.layer3labs.io/open-weights/kimi-k3-open-weights-guide",
            "https://www.microsoft.com/en-us/research/publication/rag-vs-fine-tuning-pipelines-tradeoffs-and-a-case-study-on-agriculture/",
            "https://www.mindstudio.ai/blog/kimi-k3-vs-claude-fable-5-agentic-coding",
            "https://www.reddit.com/r/LocalLLM/comments/1v2z4m2/rig_recommendation_for_a_scientist/",
            "https://www.reddit.com/r/LocalLLaMA/comments/1ouxy9y/finetune_for_rag/",
            "https://www.reddit.com/r/Rag/comments/1ouz0zs/finetune_for_rag/",
            "https://www.redhat.com/en/topics/ai/rag-vs-fine-tuning",
            "https://www.scifiniti.com/3104-4719/3/2026.0031",
            "https://www.trickywombat.ai/signals/rag-vs-fine-tuning"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 1,
          "provider": "kimi_builtin_web_search",
          "model": "kimi-k2.6",
          "observedAt": "2026-07-22T01:34:12+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 0,
          "discardedPlaceholderCount": 0,
          "citedUrls": [],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 2,
          "provider": "kimi_builtin_web_search",
          "model": "kimi-k2.6",
          "observedAt": "2026-07-22T01:34:12+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 0,
          "discardedPlaceholderCount": 0,
          "citedUrls": [],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 3,
          "provider": "kimi_builtin_web_search",
          "model": "kimi-k2.6",
          "observedAt": "2026-07-22T01:34:12+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 18,
          "discardedPlaceholderCount": 2,
          "citedUrls": [
            "https://36kr.com/",
            "https://ai.google.dev/docs/",
            "https://arxiv.org/abs/2005.11401",
            "https://arxiv.org/abs/2106.09685",
            "https://arxiv.org/abs/2309.16609",
            "https://arxiv.org/abs/2312.10997",
            "https://arxiv.org/abs/2312.11805",
            "https://arxiv.org/abs/2401.15884",
            "https://arxiv.org/abs/2404.16130",
            "https://blog.google/technology/ai/",
            "https://help.aliyun.com/zh/model-studio/user-guide/",
            "https://mp.weixin.qq.com/s/",
            "https://platform.moonshot.cn/docs/",
            "https://platform.moonshot.cn/docs/intro/kimi-llm-platform",
            "https://www.bilibili.com/",
            "https://www.jiqizhixin.com/",
            "https://www.mckinsey.com/",
            "https://zhuanlan.zhihu.com/p/"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 1,
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "observedAt": "2026-07-22T01:29:56+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 9,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://cloud.tencent.com/developer/article/2589138",
            "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
            "https://hub.baai.ac.cn/view/35723",
            "https://tech.ifeng.com/c/8uuO6XHofwJ",
            "https://www.163.com/dy/article/KO7GBD17051189P5.html",
            "https://www.thepaper.cn/newsDetail_forward_32034602",
            "https://zhuanlan.zhihu.com/p/2035991209609576779",
            "https://zhuanlan.zhihu.com/p/2060028263477540134",
            "https://zhuanlan.zhihu.com/p/2061520970595275127"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 2,
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "observedAt": "2026-07-22T01:29:56+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 9,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://cloud.tencent.com/developer/article/2589138",
            "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
            "https://hub.baai.ac.cn/view/35723",
            "https://tech.ifeng.com/c/8uuO6XHofwJ",
            "https://www.163.com/dy/article/KO7GBD17051189P5.html",
            "https://www.thepaper.cn/newsDetail_forward_32034602",
            "https://zhuanlan.zhihu.com/p/2035991209609576779",
            "https://zhuanlan.zhihu.com/p/2060028263477540134",
            "https://zhuanlan.zhihu.com/p/2061520970595275127"
          ],
          "targetCited": false,
          "brandMentioned": false
        },
        {
          "runIndex": 3,
          "provider": "qwen_dashscope_web_search",
          "model": "qwen-plus",
          "observedAt": "2026-07-22T01:29:56+00:00",
          "citationSurfaceVerifiable": true,
          "concreteCitationUrlCount": 9,
          "discardedPlaceholderCount": 0,
          "citedUrls": [
            "https://cloud.tencent.com/developer/article/2589138",
            "https://developer.cloud.tencent.com/article/2652153?policyId=1004",
            "https://hub.baai.ac.cn/view/35723",
            "https://tech.ifeng.com/c/8uuO6XHofwJ",
            "https://www.163.com/dy/article/KO7GBD17051189P5.html",
            "https://www.thepaper.cn/newsDetail_forward_32034602",
            "https://zhuanlan.zhihu.com/p/2035991209609576779",
            "https://zhuanlan.zhihu.com/p/2060028263477540134",
            "https://zhuanlan.zhihu.com/p/2061520970595275127"
          ],
          "targetCited": false,
          "brandMentioned": false
        }
      ]
    }
  ],
  "strictTrend": {
    "comparable": false,
    "partiallyComparable": true,
    "reason": "provider_model_drift",
    "baselineEvidence": {
      "datasetUrl": "https://course.dragonai.tech/datasets/ai-search-citation-benchmark-snapshot-2026-07-22.json",
      "windowId": "B"
    },
    "postpublicationEvidence": {
      "datasetUrl": "https://course.dragonai.tech/datasets/ai-search-citation-benchmark-snapshot-2026-07-22.json",
      "windowId": "C"
    },
    "providerModels": [
      {
        "provider": "gemini_google_search_grounding",
        "baselineModel": "gemini-2.5-flash",
        "currentModel": "gemini-2.5-flash",
        "comparable": true,
        "reason": "same_model"
      },
      {
        "provider": "kimi_builtin_web_search",
        "baselineModel": "moonshot-v1-128k",
        "currentModel": "kimi-k2.6",
        "comparable": false,
        "reason": "provider_missing_or_model_changed"
      },
      {
        "provider": "qwen_dashscope_web_search",
        "baselineModel": "qwen-plus",
        "currentModel": "qwen-plus",
        "comparable": true,
        "reason": "same_model"
      }
    ],
    "comparableProviders": [
      "gemini_google_search_grounding",
      "qwen_dashscope_web_search"
    ],
    "excludedProviders": [
      "kimi_builtin_web_search"
    ],
    "baselineTargetCitationRuns": 0,
    "currentTargetCitationRuns": 0,
    "targetCitationRunDelta": 0,
    "baselineVerifiableRuns": 6,
    "currentVerifiableRuns": 6,
    "baselineCitationRate": 0,
    "currentCitationRate": 0,
    "citationRateDelta": 0,
    "allProviderSnapshotCountsAsStrictTrend": false
  },
  "limitationsZh": [
    "正式基准只有一个固定中文无品牌问题、三个提供商和每窗口每家三次运行。",
    "Kimi 从 moonshot-v1-128k 迁移到 kimi-k2.6，因此不计入 B 到 C 的同模型严格趋势。",
    "C 窗口由 24 小时时间盒触发，不能归因于目标页面发布、抓取、索引或改写。",
    "引用 URL 出现不证明页面内容被答案吸收、来源正确、排名稳定或产生业务结果。",
    "模型、索引、地区、账号、搜索工具和展示机制可能变化。"
  ],
  "integrity": {
    "algorithm": "sha256",
    "canonicalPayloadSha256": "55ef770dcfbf56bcdf3d88f9ceee3de2512089719304eb27cae9e4ea59783051",
    "scope": "all_top_level_fields_except_integrity"
  },
  "title": "千问、Kimi、Gemini 固定引用问题双窗口基准（2026-07-21—22）",
  "description": "同一固定中文无品牌元问题在正式发布前基线 B 与发布后时间盒 C 的18条逐运行记录、具体引用 URL、零目标结果和模型漂移边界。",
  "url": "https://course.dragonai.tech/datasets/ai-search-citation-benchmark-snapshot-2026-07-22.json",
  "creator": "烛龙智元内容研究组"
}