{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://github.com/api-evangelist/flux/blob/main/json-schema/flux-generation-request-schema.json",
  "title": "Flux Image Generation Request",
  "description": "JSON Schema for image generation requests submitted to the Black Forest Labs Flux API. Covers the common parameters shared across FLUX model variants and the asynchronous task response format.",
  "type": "object",
  "required": ["prompt"],
  "properties": {
    "prompt": {
      "type": "string",
      "description": "Text description of the image to generate. Detailed prompts with subject, style, lighting, and composition details yield the best results.",
      "maxLength": 10000
    },
    "width": {
      "type": "integer",
      "description": "Width of the output image in pixels. Must be a multiple of 32. Used with height for standard generation endpoints.",
      "minimum": 256,
      "maximum": 1440
    },
    "height": {
      "type": "integer",
      "description": "Height of the output image in pixels. Must be a multiple of 32.",
      "minimum": 256,
      "maximum": 1440
    },
    "aspect_ratio": {
      "type": "string",
      "description": "Aspect ratio for Ultra model generation. Used instead of explicit width/height.",
      "enum": ["21:9", "16:9", "4:3", "3:2", "1:1", "2:3", "3:4", "9:16", "9:21"]
    },
    "steps": {
      "type": "integer",
      "description": "Number of diffusion inference steps. More steps increase quality but also processing time.",
      "minimum": 1,
      "maximum": 50
    },
    "guidance": {
      "type": "number",
      "description": "Classifier-free guidance scale. Higher values adhere more closely to the prompt; lower values allow more creative variation.",
      "minimum": 1.5,
      "maximum": 10.0
    },
    "seed": {
      "type": "integer",
      "description": "Random seed for reproducible generation. Omit for a random seed.",
      "minimum": 0
    },
    "safety_tolerance": {
      "type": "integer",
      "description": "Content safety filter level from 0 (most restrictive) to 6 (least restrictive). Default is 2.",
      "minimum": 0,
      "maximum": 6,
      "default": 2
    },
    "output_format": {
      "type": "string",
      "description": "Output image file format.",
      "enum": ["jpeg", "png"],
      "default": "jpeg"
    },
    "prompt_upsampling": {
      "type": "boolean",
      "description": "Whether to automatically enhance the prompt using an LLM before generation.",
      "default": false
    },
    "raw": {
      "type": "boolean",
      "description": "Ultra model only. If true, produces more natural photorealistic images without additional artistic processing.",
      "default": false
    }
  },
  "$defs": {
    "TaskResponse": {
      "type": "object",
      "title": "TaskResponse",
      "description": "Response returned immediately when a generation or editing request is submitted. The caller must poll the get_result endpoint with the returned ID.",
      "required": ["id"],
      "properties": {
        "id": {
          "type": "string",
          "format": "uuid",
          "description": "Unique task identifier. Use this to poll GET /v1/get_result?id={id}."
        },
        "polling_url": {
          "type": "string",
          "format": "uri",
          "description": "Pre-built polling URL for convenience."
        }
      }
    },
    "ResultResponse": {
      "type": "object",
      "title": "ResultResponse",
      "description": "Response from polling the generation result. When status is Ready, the result contains a download URL.",
      "required": ["id", "status"],
      "properties": {
        "id": {
          "type": "string",
          "format": "uuid",
          "description": "Task identifier matching the original request."
        },
        "status": {
          "type": "string",
          "description": "Current task status. Poll until Ready or an error state.",
          "enum": ["Pending", "Processing", "Ready", "Error", "Content Moderated", "Request Moderated"]
        },
        "result": {
          "$ref": "#/$defs/GenerationResult"
        }
      }
    },
    "GenerationResult": {
      "type": "object",
      "title": "GenerationResult",
      "description": "The successful output of a completed image generation or editing task.",
      "properties": {
        "sample": {
          "type": "string",
          "format": "uri",
          "description": "Pre-signed URL to download the generated or edited image. This URL expires after a short time."
        },
        "prompt": {
          "type": "string",
          "description": "The effective prompt used for generation, which may differ from the submitted prompt if prompt_upsampling was enabled."
        },
        "seed": {
          "type": "integer",
          "description": "The random seed that was used for this generation. Useful for reproducing the same result.",
          "minimum": 0
        }
      }
    },
    "KontextEditRequest": {
      "type": "object",
      "title": "KontextEditRequest",
      "description": "Request body for FLUX.1 Kontext image editing. Extends the base generation request with an input image field.",
      "required": ["prompt", "image"],
      "properties": {
        "prompt": {
          "type": "string",
          "description": "Text instruction describing the desired edit to apply to the input image.",
          "maxLength": 10000
        },
        "image": {
          "type": "string",
          "description": "Base64-encoded input image to edit. Supported formats are JPEG and PNG.",
          "contentEncoding": "base64",
          "contentMediaType": "image/jpeg"
        },
        "steps": {
          "type": "integer",
          "description": "Diffusion inference steps.",
          "minimum": 1,
          "maximum": 50
        },
        "guidance": {
          "type": "number",
          "description": "Guidance scale.",
          "minimum": 1.5,
          "maximum": 10.0
        },
        "seed": {
          "type": "integer",
          "description": "Random seed.",
          "minimum": 0
        },
        "safety_tolerance": {
          "type": "integer",
          "description": "Safety filter level (0-6).",
          "minimum": 0,
          "maximum": 6
        },
        "output_format": {
          "type": "string",
          "description": "Output format.",
          "enum": ["jpeg", "png"]
        }
      }
    }
  }
}