{
  "schemaVersion": 1,
  "retrievedAt": "2026-09-05T06:36:03.277Z",
  "refreshCadence": "hourly",
  "license": "CC BY 4.0",
  "canonicalPage": "https://llm-stats.com/best-ai-for-architects",
  "methodologyUrl": "https://llm-stats.com/research/best-ai-for-architects/methodology.md",
  "rankingCluster": "Places, Architecture & Environments",
  "rankingMethod": "Conservative TrueSkill (mu minus three sigma)",
  "scope": "Underlying image models for architectural visualization; not CAD, BIM, rendering, or project-management products",
  "limitations": [
    "The cluster includes interiors, buildings, landscapes, cities, and environmental scenes; it is not an architecture-only benchmark.",
    "Visual preference does not establish geometric fidelity, constructability, accessibility, code compliance, or professional fitness.",
    "Prices may omit subscriptions, editing tools, upscaling, retries, and human correction time."
  ],
  "records": [
    {
      "rank": 1,
      "modelId": "gemini-2.5-flash-image",
      "model": "Gemini 2.5 Flash Image (Nano Banana)",
      "organization": "google",
      "conservativeTrueSkillScore": 2418,
      "blindComparisons": 514,
      "winRatePercent": 30.93,
      "generationUsd": 0.04013384,
      "throughput": 9.85835684318183
    },
    {
      "rank": 2,
      "modelId": "gpt-image-1.5",
      "model": "GPT Image 1.5",
      "organization": "openai",
      "conservativeTrueSkillScore": 2417,
      "blindComparisons": 629,
      "winRatePercent": 32.75,
      "generationUsd": 0.04894816,
      "throughput": 2.1683365548038163
    },
    {
      "rank": 3,
      "modelId": "seedream-4.5",
      "model": "Seedream 4.5",
      "organization": "bytedance",
      "conservativeTrueSkillScore": 2412,
      "blindComparisons": 441,
      "winRatePercent": 21.54,
      "generationUsd": 0.04,
      "throughput": 3.5685493937965345
    },
    {
      "rank": 4,
      "modelId": "gemini-3-pro-image-preview",
      "model": "Gemini 3 Pro Image",
      "organization": "google",
      "conservativeTrueSkillScore": 2377,
      "blindComparisons": 796,
      "winRatePercent": 39.82,
      "generationUsd": 0.13018914,
      "throughput": 3.9154400864297956
    },
    {
      "rank": 5,
      "modelId": "gpt-image-2",
      "model": "GPT Image 2",
      "organization": "openai",
      "conservativeTrueSkillScore": 2325,
      "blindComparisons": 9,
      "winRatePercent": 77.78,
      "generationUsd": 0.053,
      "throughput": 3.310203935144856
    },
    {
      "rank": 6,
      "modelId": "flux-2-max",
      "model": "Flux 2 Max",
      "organization": "black-forest-labs",
      "conservativeTrueSkillScore": 2251,
      "blindComparisons": 83,
      "winRatePercent": 18.07,
      "generationUsd": 0.06236776,
      "throughput": 1.8288131140399249
    },
    {
      "rank": 7,
      "modelId": "riverflow-2.0-pro",
      "model": "Riverflow 2.0 Pro",
      "organization": "sourceful",
      "conservativeTrueSkillScore": 2235,
      "blindComparisons": 8,
      "winRatePercent": 62.5,
      "generationUsd": 0.15,
      "throughput": 1.2843341721139785
    },
    {
      "rank": 8,
      "modelId": "flux-2-pro",
      "model": "Flux 2 Pro",
      "organization": "black-forest-labs",
      "conservativeTrueSkillScore": 2231,
      "blindComparisons": 111,
      "winRatePercent": 23.42,
      "generationUsd": 0.02,
      "throughput": 5.578773840849933
    },
    {
      "rank": 9,
      "modelId": "seedream-4",
      "model": "Seedream 4",
      "organization": "bytedance",
      "conservativeTrueSkillScore": 2213,
      "blindComparisons": 71,
      "winRatePercent": 25.35,
      "generationUsd": 0.03,
      "throughput": 5.534696679794649
    },
    {
      "rank": 10,
      "modelId": "grok-imagine-image",
      "model": "Grok Imagine Image",
      "organization": "xai",
      "conservativeTrueSkillScore": 2208,
      "blindComparisons": 41,
      "winRatePercent": 26.83,
      "generationUsd": 0.02,
      "throughput": 10.153895315430496
    }
  ]
}