{
  "docId": "019dd923-5e88-73ef-bd5d-92cb9e5864da",
  "docSlug": "dfd94aead7429e9f",
  "documentTitle": "2025 Benedict Evans 2025 Autum AI eats the world",
  "authorId": "BenedictEvans",
  "authorName": "Benedict Evans",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "independent_analyst",
  "sourceTypeLabel": "Independent analyst",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.778,
  "pageNumber": 35,
  "pageCount": 90,
  "prevPage": 34,
  "nextPage": 36,
  "slideType": "industry_trends",
  "function": "compare_peers",
  "density": "dense",
  "nDataPoints": 20,
  "notes": "The y-axis is truncated starting at 90%.",
  "elementsJson": [
    "headline_text",
    "subtitle_text",
    "bar_chart_vertical",
    "footnote"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-92cb9e5864da/35",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-92cb9e5864da",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-92cb9e5864da.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-92cb9e5864da#slide-35",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Dozens of (saturated) benchmarks to choose from, but on the most general, the leaders are very close",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-4e52-728e-927b-ab8cb3eb5333",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.28,
        "x": 0.205,
        "y": 0.35
      },
      "kind": "chart",
      "text": "LMArena Text performance comparison",
      "attrs": null,
      "subkind": "bar-vertical",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "1a0d4e8a-a67d-46a4-8813-60d69d6b5cf3",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.28,
        "x": 0.55,
        "y": 0.35
      },
      "kind": "chart",
      "text": "MMLU-Pro performance comparison",
      "attrs": null,
      "subkind": "bar-vertical",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "9e3c16bf-f074-41b1-ac26-4948b328c490",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.1,
        "x": 0.05,
        "y": 0.35
      },
      "kind": "legend",
      "text": "Anthropic, Google, OpenAI, Other: China, Other: Western",
      "attrs": null,
      "subkind": null,
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d6f0709c-f144-4028-8494-4588faff8696",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Relative performance score: 100%",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-4e52-728e-927b-ae6f3102c446",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.02,
        "w": 0.2,
        "x": 0.205,
        "y": 0.91
      },
      "kind": "source-note",
      "text": "Source: LMArena, Artificial Analysis",
      "attrs": null,
      "subkind": null,
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "67af3b10-d81c-4e00-a2c1-8636636fe75e",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.59,
        "x": 0.205,
        "y": 0.05
      },
      "kind": "title",
      "text": "Models converge and leaders change weekly",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "59fb1d4a-3aa2-4cfd-9042-bc8f539a4320",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.03,
        "w": 0.59,
        "x": 0.205,
        "y": 0.11
      },
      "kind": "title",
      "text": "Dozens of (saturated) benchmarks to choose from, but on the most general, the leaders are very close",
      "attrs": null,
      "subkind": "subtitle",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "7274eee6-aaa2-4ab3-9d32-5c9c43ef8141",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-1b83-7130-9342-63c9e2a2ee47",
      "evidence": "Title 'Models converge and leaders change weekly'",
      "confidence": 90
    },
    {
      "name": "Chart Selection Guide",
      "slug": "chart-selection-guide",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "fa477415-caba-4c1e-9d2e-446a9c8ae9a3",
      "evidence": "The slide features multiple charts (bar and scatter), indicating a consideration of chart selection.",
      "confidence": 0.7
    },
    {
      "name": "Lie Factor",
      "slug": "lie-factor",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "019dd95a-1b83-7130-9342-6a3738df59a8",
      "evidence": "Truncated y-axis flagged 'NB truncated axis'",
      "confidence": 60
    },
    {
      "name": "Small Multiples",
      "slug": "small-multiples",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "019dd95a-1b83-7130-9342-6432f9b567cd",
      "evidence": "Two side-by-side benchmark charts (LMArena, MMLU-Pro)",
      "confidence": 90
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 48,
      "from": 14,
      "beatId": "019dd95a-07a7-7579-bd65-4f5b40875a96",
      "arcName": "The Mountain",
      "arcSlug": "mountain",
      "beatName": "Rising Action",
      "beatSlug": "mountain-rising-action",
      "evidence": "Inside tech: $400bn capex, $1.4tr OpenAI, models commoditize, value capture unclear",
      "position": 2,
      "confidence": 70,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    },
    {
      "to": 48,
      "from": 1,
      "beatId": "019dd95a-07a7-7579-bd65-5b209794247f",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Platform shift evidence + capex surge + model proliferation data",
      "position": 1,
      "confidence": 60,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    }
  ],
  "loops": [
    {
      "to": 48,
      "from": 33,
      "name": "Aha Moment",
      "slug": "03-aha-moment",
      "bestFor": "Data-heavy sections, research findings, analytical arguments",
      "matchId": "019dd95a-08f8-7619-ac2e-c68666bae910",
      "evidence": "Model convergence (35), 5% paying (37), 'where is value capture?' (48)",
      "position": 3,
      "objective": "Reveal that despite spend, value capture location is unclear",
      "structure": "The Problem/Question -> What the Data Shows -> The Insight",
      "confidence": 75,
      "description": "Establish a tension, present data, then deliver the insight that changes everything"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}