# NVIDIA Vision Model Configuration
# This file manages API keys and model fallback preferences

# =============================================================================
# API CONFIGURATION
# =============================================================================
# For production, use environment variables instead of hardcoding keys
# Set these in your .env file or system environment

NVIDIA_API_KEY=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
GEMINI_API_KEY=AIzaSyDkiYr-eSkqIXpZ1fHlik_YFsFtfQoFi0w

# =============================================================================
# MODEL CONFIGURATION
# =============================================================================
# Models are tried in order. If one fails, the next is attempted.
# Format: model_name (comma-separated, in priority order)

# NVIDIA Vision Models (in order of preference)
# These models support image analysis with vision capabilities
NVIDIA_VISION_MODELS=meta/llama-3.2-90b-vision-instruct,microsoft/phi-3-vision-128k-instruct,meta/llama-3.2-11b-vision-instruct,nvidia/neva-22b

# NVIDIA Text Models (for text-only analysis as additional fallback)
# These are faster and work for non-image tasks
NVIDIA_TEXT_MODELS=meta/llama-3.1-405b-instruct,meta/llama-3.1-70b-instruct,meta/llama-3.1-8b-instruct,mistralai/mixtral-8x7b-instruct-v0.1,mistralai/mistral-7b-instruct-v0.3

# Combined NVIDIA Models (vision + text for maximum coverage)
NVIDIA_MODELS=meta/llama-3.2-90b-vision-instruct,microsoft/phi-3-vision-128k-instruct,meta/llama-3.2-11b-vision-instruct,nvidia/neva-22b,meta/llama-3.1-405b-instruct,meta/llama-3.1-70b-instruct,meta/llama-3.1-8b-instruct,mistralai/mixtral-8x7b-instruct-v0.1

# Gemini Models (for text analysis fallback)
GEMINI_MODELS=gemini-2.0-flash-exp,gemini-2.0-flash-thinking-exp-1219,gemini-exp-1206,gemini-2.0-flash,gemini-1.5-flash,gemini-1.5-flash-8b,gemini-1.5-pro

# =============================================================================
# REQUEST CONFIGURATION
# =============================================================================
# Maximum tokens for model responses
MAX_TOKENS=500

# Temperature for model creativity (0.0 = deterministic, 1.0 = creative)
TEMPERATURE=0.2

# Request timeout in seconds
REQUEST_TIMEOUT=30

# Maximum retry attempts per model
MAX_RETRIES=2

# =============================================================================
# FEATURE FLAGS
# =============================================================================
# Enable/disable streaming responses
ENABLE_STREAMING=true

# Enable verbose logging
VERBOSE_LOGGING=true

# Enable fallback to Gemini if all NVIDIA models fail
ENABLE_GEMINI_FALLBACK=true