{
  "listings": [
    {
      "slug": "adobe-firefly",
      "name": "Adobe Firefly API",
      "company": "Adobe Inc.",
      "grade": "D",
      "fit": "caveat",
      "why": "Reference blobs (512 to 2048 pixels, 10 MB) and Firefly Custom Models give two routes, but no identity-preservation claim or retirement policy was found.",
      "source": "https://raw.githubusercontent.com/AdobeDocs/ffs-firefly-api/main/static/firefly-api.json",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/adobe-firefly"
    },
    {
      "slug": "black-forest-labs",
      "name": "Black Forest Labs FLUX API",
      "company": "Black Forest Labs Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "Up to ten references on FLUX 3 and eight on FLUX.2, a character-consistency guide and finetune endpoints give two routes, but the guide is vendor examples with no recognition rate, and FLUX 3 can't yet be pinned to a dated version.",
      "source": "https://docs.bfl.ml/guides/usecases_editing_character_consistency.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/black-forest-labs"
    },
    {
      "slug": "bria",
      "name": "Bria API",
      "company": "Bria Artificial Intelligence Ltd.",
      "grade": "BB",
      "fit": "caveat",
      "why": "Edits take up to four reference images, but no identity-preservation claim or fine-tune route was found, and there is no deprecation policy to keep a model callable.",
      "source": "https://docs.bria.ai/image-editing/editing/edit-image.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/bria"
    },
    {
      "slug": "fal-image",
      "name": "fal image models",
      "company": "Features \u0026 Labels, Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "Multi-reference edit models and LoRA training endpoints exist, but acc-28 found only 1 or 2 of 40 portraits recognisable after eight flux-lora-fast-training runs, and deprecated endpoints carry no removal date.",
      "source": "https://github.com/fal-ai/fal/issues/1024",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/fal-image"
    },
    {
      "slug": "flora",
      "name": "FLORA",
      "company": "FLORA AND FAUNA FLOWERSHOP, INC.",
      "grade": "B",
      "fit": "caveat",
      "why": "Ideogram Character and LoRA Style training through the API give reference and fine-tune routes, but no recognition rate or model retirement policy is published.",
      "source": "https://docs.flora.ai/plans-and-billing/model-pricing.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/flora"
    },
    {
      "slug": "ideogram",
      "name": "Ideogram API",
      "company": "Ideogram AI, Inc.",
      "grade": "D",
      "fit": "caveat",
      "why": "ideogram-3-character takes character_reference_images with an optional mask, and custom-model endpoints exist, but it is the 3.0 line, no recognition rate is given, and models change without a dated changelog.",
      "source": "https://developer.ideogram.ai/v2/api-reference/images/generate/ideogram-3-character.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/ideogram"
    },
    {
      "slug": "leonardo-ai",
      "name": "Leonardo.Ai API",
      "company": "Leonardo Interactive Pty Ltd",
      "grade": "F",
      "fit": "caveat",
      "why": "Character and style reference guidances exist on Lucid models, but no reference count or identity claim was read, and the deprecations page shows models removed with little or no notice.",
      "source": "https://docs.leonardo.ai/reference/creategeneration.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/leonardo-ai"
    },
    {
      "slug": "openai-image-api",
      "name": "OpenAI Image API",
      "company": "OpenAI OpCo, LLC",
      "grade": "BB",
      "fit": "caveat",
      "why": "Edits take several reference images at high fidelity and image models get three months' notice, but the guide says recurring characters may not stay consistent, there is no fine-tune for image models, and acc-08 lost continuity across pages.",
      "source": "https://developers.openai.com/api/docs/guides/image-generation",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/openai-image-api"
    },
    {
      "slug": "qwen-image-2-1",
      "name": "Qwen-Image 2.1 (Model Studio)",
      "company": "Alibaba Cloud (Singapore) Private Limited",
      "grade": "B",
      "fit": "caveat",
      "why": "Up to 10 reference images and identity preservation for people are model-card claims with no measured rate, and the API's model lifecycle wasn't readable.",
      "source": "https://huggingface.co/Qwen/Qwen-Image-2.1",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/qwen-image-2-1"
    },
    {
      "slug": "replicate-image",
      "name": "Replicate image models",
      "company": "Replicate, LLC",
      "grade": "B",
      "fit": "caveat",
      "why": "The docs index lists a FLUX fine-tune guide and versioned LoRA models, an identity route that can be pinned by version, but no recognition rate is given and no retirement notice applies to official models.",
      "source": "https://replicate.com/docs/llms.txt",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/replicate-image"
    },
    {
      "slug": "seedream-on-byteplus-modelark",
      "name": "Seedream on BytePlus ModelArk",
      "company": "BytePlus Pte. Ltd.",
      "grade": "C",
      "fit": "caveat",
      "why": "References plus generated images can total 15 in a sequential request, which suits a set of related images, but there is no fine-tune route or identity claim, and models carry dated version ids that get deactivated.",
      "source": "https://docs.byteplus.com/en/docs/modelark/image-generation-api",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/seedream-on-byteplus-modelark"
    },
    {
      "slug": "stability-ai-image",
      "name": "Stability AI Image API",
      "company": "Stability AI Ltd",
      "grade": "F",
      "fit": "caveat",
      "why": "Style guide and style transfer take a reference image, but there is no identity reference, no fine-tune in the API and no retirement notice period.",
      "source": "https://docs.stability.ai/llms.txt",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/stability-ai-image"
    },
    {
      "slug": "iris-3b",
      "name": "Iris-3B",
      "company": "Speridlabs",
      "grade": "C",
      "fit": "no",
      "why": "No reference-image input and no fine-tune guide in the card or README.",
      "source": "https://huggingface.co/speridlabs/iris-3b",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/iris-3b"
    },
    {
      "slug": "recraft",
      "name": "Recraft API",
      "company": "Recraft Inc.",
      "grade": "D",
      "fit": "no",
      "why": "Reference images (1 to 10) define a style, not a subject, and no fine-tune or identity reference is offered in the API.",
      "source": "https://www.recraft.ai/docs/api-reference/agents.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/recraft"
    },
    {
      "slug": "surgepix",
      "name": "SurgePix API",
      "company": "SurgePix",
      "grade": "F",
      "fit": "no",
      "why": "No reference-image input or fine-tune is documented for generation skills.",
      "source": "https://surgepix.ai/blog/developer-guides/call-skills",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/surgepix"
    }
  ],
  "marks": "yes: the docs show it does the job as the success test says. caveat: it does the job with a limit an agent should know. no: it doesn't do this job. unchecked: we couldn't read it.",
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "attributionRequired": true,
    "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-11-16",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.8",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run",
    "terms": "https://www.anchorterminal.com/terms/#data"
  },
  "useCase": {
    "accounts": 3,
    "after": [
      "video-generation.image-to-video",
      "image-generation.project-asset-batch"
    ],
    "area": "content",
    "asks": [
      "same character on every page",
      "headshots that look like me",
      "keep the style of my earlier cards",
      "use these references"
    ],
    "before": [
      "storage.download"
    ],
    "blockers": [
      "model that defined the style is retired",
      "fine-tune preserves clothing and lighting but not identity",
      "continuity lost over long multi-image projects"
    ],
    "capabilities": [
      "image.reference",
      "image.edit",
      "image.generate"
    ],
    "category": "image-generation",
    "categoryName": "Image generation",
    "checked": "2026-11-15",
    "companies": {
      "fit": 0,
      "fitWithLimit": 12,
      "tagged": 45
    },
    "competition": "none",
    "fits": null,
    "fitsWithLimit": [
      "adobe-firefly",
      "black-forest-labs",
      "bria",
      "fal-image",
      "flora",
      "ideogram",
      "leonardo-ai",
      "openai-image-api",
      "qwen-image-2-1",
      "replicate-image",
      "seedream-on-byteplus-modelark",
      "stability-ai-image"
    ],
    "id": "image-generation.character-consistency",
    "listings": {
      "caveat": 12,
      "no": 3,
      "notRead": 0,
      "tagged": 46,
      "unchecked": 0,
      "yes": 0
    },
    "name": "Keep a character, person or style the same across many images",
    "status": "draft",
    "success": "The same subject kept across a set of images, using reference images or a fine-tune the API accepts, with the same model still callable next month. Measured: across 10 images a reviewer recognises the same subject in at least 9.",
    "url": "https://www.anchorterminal.com/api/v1/use-cases/image-generation.character-consistency.json",
    "weight": 6,
    "weightBasis": "3 of 37 accounts in this category (game style lost to deprecation, manga continuity, LoRA headshots at 1 or 2 of 40); 11 video-generation accounts show the same need upstream of image-to-video"
  }
}
