Model Catalog

Overview

The model catalog provides structured metadata for all available AI models. Each model definition includes:

Model ID and display name
API endpoint
Supported input parameters with types and constraints
Default values and validation rules

Model Arrays

Text-to-Image Models

import { t2iModels, getModelById } from './lib/models.js';

const model = getModelById('flux-dev');

Array: t2iModels
Count: 50+ models including Flux, Midjourney, DALL-E, Stable Diffusion variants

Text-to-Video Models

import { t2vModels, getVideoModelById } from './lib/models.js';

const model = getVideoModelById('kling-v2.6-pro-t2v');

Array: t2vModels
Count: 30+ models including Kling, Runway, Sora, Luma, Hailuo

Image-to-Image Models

import { i2iModels, getI2IModelById } from './lib/models.js';

const model = getI2IModelById('flux-kontext-dev-i2i');

Array: i2iModels
Count: 60+ models including editing tools, upscalers, style transfer

Image-to-Video Models

import { i2vModels, getI2VModelById } from './lib/models.js';

const model = getI2VModelById('runway-image-to-video');

Array: i2vModels
Count: 15+ models for animating still images

Video-to-Video Models

import { v2vModels, getV2VModelById } from './lib/models.js';

const model = getV2VModelById('video-watermark-remover');

Array: v2vModels
Count: Processing tools like watermark removal

Model Object Schema

Basic Structure

{
  id: string;              // Unique identifier
  name: string;            // Display name
  endpoint?: string;       // API endpoint (defaults to id)
  family?: string;         // Model family grouping
  inputs: {                // Input parameter definitions
    [key: string]: {
      type: string;        // 'string' | 'int' | 'boolean' | 'array'
      title: string;       // Display label
      name: string;        // Parameter name
      description: string; // Help text
      default?: any;       // Default value
      enum?: any[];        // Allowed values
      minValue?: number;   // Min constraint
      maxValue?: number;   // Max constraint
      step?: number;       // Increment step
      examples?: any[];    // Example values
    }
  }
}

Example: Flux Dev Model

{
  "id": "flux-dev",
  "name": "Flux Dev",
  "endpoint": "flux-dev-image",
  "inputs": {
    "prompt": {
      "type": "string",
      "title": "Prompt",
      "name": "prompt",
      "description": "Text prompt describing the image. Length: 2-3000 characters.",
      "examples": [
        "Extreme close-up of a single tiger eye..."
      ]
    },
    "width": {
      "type": "int",
      "title": "Width",
      "name": "width",
      "description": "Width divisible by 64",
      "default": 1024,
      "minValue": 128,
      "maxValue": 2048,
      "step": 64
    },
    "num_images": {
      "type": "int",
      "title": "Number of images",
      "name": "num_images",
      "default": 1,
      "minValue": 1,
      "maxValue": 4
    }
  }
}

I2I Model Extensions

Image-to-Image models have additional fields:

{
  imageField: string;      // Image input field name
  hasPrompt: boolean;      // Whether prompt is supported
  maxImages?: number;      // Max reference images (default 1)
}

Example: Flux Kontext Dev I2I

{
  "id": "flux-kontext-dev-i2i",
  "name": "Flux Kontext Dev I2I",
  "endpoint": "flux-kontext-dev-i2i",
  "family": "kontext",
  "imageField": "images_list",
  "hasPrompt": true,
  "maxImages": 10,
  "inputs": {
    "prompt": { /* ... */ },
    "aspect_ratio": {
      "enum": ["16:9", "9:16", "1:1", "4:3", "3:4", "3:2", "2:3", "21:9", "9:21"],
      "default": "1:1"
    },
    "num_images": {
      "default": 1,
      "minValue": 1,
      "maxValue": 4
    }
  }
}

The imageField tells the client which payload key to use for images:

image_url - Single image URL
images_list - Array of image URLs
model_image_url - Model/reference image

I2V Model Extensions

Image-to-Video models have:

{
  imageField: string;      // Image input field name
  hasPrompt: boolean;      // Whether prompt is supported
}

Helper Functions

Get Model by ID

import { getModelById, getVideoModelById, getI2IModelById, getI2VModelById } from './lib/models.js';

const t2iModel = getModelById('flux-dev');
const t2vModel = getVideoModelById('kling-v2.6-pro-t2v');
const i2iModel = getI2IModelById('flux-kontext-dev-i2i');
const i2vModel = getI2VModelById('runway-image-to-video');

string

required

The model ID to look up

model

object | undefined

Model definition object or undefined if not found

Get Aspect Ratios

import { 
  getAspectRatiosForModel,
  getAspectRatiosForVideoModel,
  getAspectRatiosForI2IModel,
  getAspectRatiosForI2VModel 
} from './lib/models.js';

const ratios = getAspectRatiosForModel('flux-dev');
// Returns: ['16:9', '9:16', '1:1', '4:3', '3:2', '21:9']

ratios

array

Array of supported aspect ratio strings

Defaults if not defined:

T2I: ['1:1', '16:9', '9:16', '4:3', '3:2', '21:9']
T2V/I2V: ['16:9', '9:16', '1:1']
I2I: ['1:1', '16:9', '9:16']

Get Durations

import { getDurationsForModel, getDurationsForI2VModel } from './lib/models.js';

const durations = getDurationsForModel('kling-v2.6-pro-t2v');
// Returns: [5, 10]

const i2vDurations = getDurationsForI2VModel('runway-image-to-video');

durations

array

Array of supported duration values in seconds

Logic:

If inputs.duration.enum exists, return it
If inputs.duration.minValue/maxValue/step exist, generate range
If inputs.duration.default exists, return [default]
Otherwise return []

Get Resolutions

import { 
  getResolutionsForModel,
  getResolutionsForVideoModel,
  getResolutionsForI2IModel,
  getResolutionsForI2VModel 
} from './lib/models.js';

const resolutions = getResolutionsForVideoModel('runway-text-to-video');
// Returns: ['720p', '1080p']

const imageRes = getResolutionsForModel('bytedance-seedream-v4');
// Returns: ['1K', '2K', '4K']

resolutions

array

Array of resolution/quality options

Some models use quality instead of resolution (e.g., basic, high). These functions check both fields.

Get Quality Field Name

import { getQualityFieldForModel, getQualityFieldForI2IModel } from './lib/models.js';

const field = getQualityFieldForModel('flux-dev');
// Returns: null (no quality/resolution field)

const field2 = getQualityFieldForModel('bytedance-seedream-v4');
// Returns: 'resolution'

const field3 = getQualityFieldForModel('seedance-v2.0-t2v');
// Returns: 'quality'

fieldName

string | null

The payload field name ('resolution', 'quality', or null)

Get Max Images for I2I

import { getMaxImagesForI2IModel } from './lib/models.js';

const max = getMaxImagesForI2IModel('flux-kontext-dev-i2i');
// Returns: 10

const max2 = getMaxImagesForI2IModel('gpt4o-image-to-image');
// Returns: 5

maxImages

number

Maximum number of reference images (defaults to 1)

Accessing Model Data

List All Models

import { t2iModels } from './lib/models.js';

t2iModels.forEach(model => {
  console.log(`${model.name} (${model.id})`);
});

Check If Model Has Prompt

const model = getI2IModelById('ai-image-upscaler');
if (model.hasPrompt) {
  // Show prompt input
} else {
  // Hide prompt input
}

Get Input Constraints

const model = getModelById('flux-dev');
const widthInput = model.inputs.width;

console.log(`Width range: ${widthInput.minValue} - ${widthInput.maxValue}`);
console.log(`Step: ${widthInput.step}`);
console.log(`Default: ${widthInput.default}`);
// Output:
// Width range: 128 - 2048
// Step: 64
// Default: 1024

Build Dynamic UI

const model = getModelById('midjourney-v7-text-to-image');

// Create select for aspect ratio
const aspectRatios = model.inputs.aspect_ratio.enum;
const select = document.createElement('select');
aspectRatios.forEach(ratio => {
  const option = document.createElement('option');
  option.value = ratio;
  option.text = ratio;
  if (ratio === model.inputs.aspect_ratio.default) {
    option.selected = true;
  }
  select.appendChild(option);
});

Common Input Parameters

Prompt (T2I/T2V/I2I/I2V)

prompt: {
  type: 'string',
  description: 'Text describing the desired output',
  minLength?: 2,
  maxLength?: 3000
}

Aspect Ratio

aspect_ratio: {
  type: 'string',
  enum: ['1:1', '16:9', '9:16', '4:3', '3:4', '21:9', ...],
  default: '1:1' | '16:9'
}

Resolution/Quality

resolution: {
  type: 'string',
  enum: ['480p', '720p', '1080p'] | ['1k', '2k', '4k'],
  default: '720p' | '1k'
}

quality: {
  type: 'string',
  enum: ['basic', 'high', 'medium'],
  default: 'basic'
}

Duration (Video)

duration: {
  type: 'int',
  enum: [5, 10, 15] | { minValue: 3, maxValue: 15, step: 1 },
  default: 5
}

Number of Images

num_images: {
  type: 'int',
  default: 1,
  minValue: 1,
  maxValue: 4
}

Image Dimensions (T2I)

width: {
  type: 'int',
  default: 1024,
  minValue: 256,
  maxValue: 2048,
  step: 64
}

height: {
  type: 'int',
  default: 1024,
  minValue: 256,
  maxValue: 2048,
  step: 64
}

Model Families

Models are grouped by family for UI organization:

flux - Flux models (Dev, Pro, Schnell, etc.)
kontext - Flux Kontext variants
midjourney - Midjourney v7 models
seedream - ByteDance Seedream series
kling - Kling video models
runway - Runway Gen-3
tools - Utility models (upscale, remove BG, etc.)
nano - Nano Banana models
gpt - OpenAI DALL-E models
ideogram - Ideogram models

Example: Filter by Family

const fluxModels = t2iModels.filter(m => m.family === 'flux');
const tools = i2iModels.filter(m => m.family === 'tools');

Validation Example

import { getModelById } from './lib/models.js';

function validateParams(modelId, params) {
  const model = getModelById(modelId);
  if (!model) throw new Error(`Model ${modelId} not found`);
  
  const errors = [];
  
  // Check width constraint
  const widthDef = model.inputs.width;
  if (widthDef) {
    if (params.width < widthDef.minValue || params.width > widthDef.maxValue) {
      errors.push(`Width must be between ${widthDef.minValue} and ${widthDef.maxValue}`);
    }
    if (params.width % widthDef.step !== 0) {
      errors.push(`Width must be divisible by ${widthDef.step}`);
    }
  }
  
  // Check aspect ratio
  const arDef = model.inputs.aspect_ratio;
  if (arDef && arDef.enum && !arDef.enum.includes(params.aspect_ratio)) {
    errors.push(`Invalid aspect ratio. Must be one of: ${arDef.enum.join(', ')}`);
  }
  
  return errors;
}

Complete Example

import { 
  getModelById, 
  getAspectRatiosForModel,
  getResolutionsForModel,
  getQualityFieldForModel 
} from './lib/models.js';

const modelId = 'bytedance-seedream-v4';
const model = getModelById(modelId);

console.log('Name:', model.name);
console.log('Endpoint:', model.endpoint || model.id);

const aspectRatios = getAspectRatiosForModel(modelId);
console.log('Aspect Ratios:', aspectRatios);

const resolutions = getResolutionsForModel(modelId);
const qualityField = getQualityFieldForModel(modelId);
console.log(`${qualityField}:`, resolutions);
// Output: resolution: ['1K', '2K', '4K']

// Build payload
const payload = {
  prompt: 'A magical forest scene',
  aspect_ratio: aspectRatios[0],
  [qualityField]: resolutions[2] // '4K'
};

API Client

Models

Overview

Model Arrays

Text-to-Image Models

Text-to-Video Models

Image-to-Image Models

Image-to-Video Models

Video-to-Video Models

Model Object Schema

Basic Structure

Example: Flux Dev Model

I2I Model Extensions

Example: Flux Kontext Dev I2I

I2V Model Extensions

Helper Functions

Get Model by ID

Get Aspect Ratios

Get Durations

Get Resolutions

Get Quality Field Name

Get Max Images for I2I

Accessing Model Data

List All Models

Check If Model Has Prompt

Get Input Constraints

Build Dynamic UI

Common Input Parameters

Prompt (T2I/T2V/I2I/I2V)

Aspect Ratio

Resolution/Quality

Duration (Video)

Number of Images

Image Dimensions (T2I)

Model Families

Example: Filter by Family

Validation Example

Complete Example

Build docs developers (and LLMs) love

API Client

Models

​Overview

​Model Arrays

​Text-to-Image Models

​Text-to-Video Models

​Image-to-Image Models

​Image-to-Video Models

​Video-to-Video Models

​Model Object Schema

​Basic Structure

​Example: Flux Dev Model

​I2I Model Extensions

​Example: Flux Kontext Dev I2I

​I2V Model Extensions

​Helper Functions

​Get Model by ID

​Get Aspect Ratios

​Get Durations

​Get Resolutions

​Get Quality Field Name

​Get Max Images for I2I

​Accessing Model Data

​List All Models

​Check If Model Has Prompt

​Get Input Constraints

​Build Dynamic UI

​Common Input Parameters

​Prompt (T2I/T2V/I2I/I2V)

​Aspect Ratio

​Resolution/Quality

​Duration (Video)

​Number of Images

​Image Dimensions (T2I)

​Model Families

​Example: Filter by Family

​Validation Example

​Complete Example

Build docs developers (and LLMs) love

Overview

Model Arrays

Text-to-Image Models

Text-to-Video Models

Image-to-Image Models

Image-to-Video Models

Video-to-Video Models

Model Object Schema

Basic Structure

Example: Flux Dev Model

I2I Model Extensions

Example: Flux Kontext Dev I2I

I2V Model Extensions

Helper Functions

Get Model by ID

Get Aspect Ratios

Get Durations

Get Resolutions

Get Quality Field Name

Get Max Images for I2I

Accessing Model Data

List All Models

Check If Model Has Prompt

Get Input Constraints

Build Dynamic UI

Common Input Parameters

Prompt (T2I/T2V/I2I/I2V)

Aspect Ratio

Resolution/Quality

Duration (Video)

Number of Images

Image Dimensions (T2I)

Model Families

Example: Filter by Family

Validation Example

Complete Example