"""
Constants for imagen-mcp providers.
This module defines provider-specific constants including supported sizes,
aspect ratios, and model identifiers.
"""
# ============================
# OpenAI GPT-Image-1 Constants
# ============================
OPENAI_API_BASE_URL = "https://api.openai.com/v1"
# OpenAI only supports these 3 sizes (verified from API)
OPENAI_SIZES = [
"1024x1024", # Square
"1024x1536", # Portrait
"1536x1024", # Landscape
]
OPENAI_MODELS = {
# Image generation models
"gpt-image-1": "gpt-image-1", # Dedicated image model (April 2025)
"gpt-5-image": "gpt-5-image", # GPT-5 with image generation (Oct 2025)
# Conversation orchestration models
"gpt-5.1": "gpt-5.1", # Latest reasoning model (Nov 2025)
"gpt-4o": "gpt-4o", # Multimodal model
}
# Default OpenAI model for image generation
DEFAULT_OPENAI_IMAGE_MODEL = "gpt-image-1"
# ============================
# Google Gemini Constants
# ============================
# Gemini model identifiers
GEMINI_MODELS = {
# Nano Banana Pro - highest quality, best for production
"gemini-3-pro-image-preview": "gemini-3-pro-image-preview",
# Gemini 2.0 Flash - fast experimental image generation
"gemini-2.0-flash-exp-image-generation": "gemini-2.0-flash-exp-image-generation",
# Imagen 3.0 - alternative image model
"imagen-3.0-generate-002": "imagen-3.0-generate-002",
# For prompt enhancement (text only)
"gemini-flash-latest": "gemini-flash-latest",
}
# Default Gemini model for image generation
DEFAULT_GEMINI_IMAGE_MODEL = "gemini-3-pro-image-preview"
# Gemini supports 10 aspect ratios
GEMINI_ASPECT_RATIOS = [
"1:1", # Square
"2:3", # Portrait (phone)
"3:2", # Landscape (photo)
"3:4", # Portrait (social)
"4:3", # Landscape (classic)
"4:5", # Portrait (Instagram)
"5:4", # Landscape (social)
"9:16", # Portrait (Stories/Reels)
"16:9", # Landscape (video)
"21:9", # Ultra-wide
]
# Gemini resolution options (use uppercase K)
GEMINI_SIZES = [
"1K", # Fast generation
"2K", # High quality (default)
"4K", # Maximum resolution
]
# Reference image limits
GEMINI_MAX_REFERENCE_IMAGES = 14
GEMINI_MAX_OBJECT_IMAGES = 6
GEMINI_MAX_HUMAN_IMAGES = 5
# ============================
# Shared Constants
# ============================
MAX_PROMPT_LENGTH = 4000 # Shared limit (OpenAI: 4000, Gemini: 8192)
MAX_RETRIES = 3
DEFAULT_TIMEOUT = 120 # seconds
# ============================
# Provider Selection Keywords
# ============================
# Keywords that suggest OpenAI is better (text-heavy tasks)
OPENAI_PREFERRED_KEYWORDS = [
"text",
"label",
"menu",
"infographic",
"diagram",
"comic",
"dialogue",
"speech bubble",
"caption",
"title",
"headline",
"poster",
"flyer",
"certificate",
"badge",
"logo with text",
"banner",
"sign",
"watermark",
]
# Keywords that suggest Gemini is better
GEMINI_PREFERRED_KEYWORDS = [
"portrait",
"headshot",
"photo",
"photorealistic",
"realistic",
"product shot",
"product photography",
"studio lighting",
"4k",
"high resolution",
"character consistency",
"reference image",
"weather",
"stock",
"current",
"today",
"real-time",
"selfie",
"person",
"face",
"beauty",
"fashion",
"magazine",
]
# Keywords that require Gemini (real-time data)
GEMINI_REQUIRED_KEYWORDS = [
"current weather",
"today's",
"real-time",
"stock price",
"latest news",
"live",
]
