// **Observações sobre as mudanças:**
// * **`RATE_LIMITS`:** Atualizado para incluir os novos modelos e remover os descontinuados. Os valores de `rpm` e `tpm` para os novos modelos são exemplos baseados em modelos semelhantes e podem precisar de ajuste fino com base na documentação exata da Groq.
import { GroqModel, ModelCapability } from '../types/groq-types.js';
import modelsConfig from './models.json' with { type: 'json' };
// Rate limits based on Groq API documentation
export const RATE_LIMITS = {
// Text completion models
'llama-3.1-8b-instant': { rpm: 30, tpm: 30000 },
'llama-3.3-70b-versatile': { rpm: 30, tpm: 6000 },
'deepseek-r1-distill-llama-70b': { rpm: 20, tpm: 6000 },
'qwen-qwq-32b': { rpm: 30, tpm: 6000 },
'qwen/qwen3-32b': { rpm: 30, tpm: 6000 },
'compound-beta': { rpm: 20, tpm: 4000 },
'compound-beta-mini': { rpm: 40, tpm: 8000 }, // Exemplo: ajustado para mini
'allam-2-7b': { rpm: 35, tpm: 25000 }, // Exemplo
'gemma2-9b-it': { rpm: 30, tpm: 15000 }, // Exemplo
'llama3-70b-8192': { rpm: 30, tpm: 6000 }, // Exemplo
'llama3-8b-8192': { rpm: 30, tpm: 30000 }, // Exemplo
'meta-llama/llama-guard-4-12b': { rpm: 30, tpm: 10000 }, // Exemplo
'meta-llama/llama-prompt-guard-2-22m': { rpm: 50, tpm: 20000 }, // Exemplo
'meta-llama/llama-prompt-guard-2-86m': { rpm: 40, tpm: 15000 }, // Exemplo
'mistral-saba-24b': { rpm: 30, tpm: 10000 }, // Exemplo
// Audio models
'whisper-large-v3': { rpm: 20, tpm: 0 },
'whisper-large-v3-turbo': { rpm: 30, tpm: 0 },
'distil-whisper-large-v3-en': { rpm: 40, tpm: 0 }, // Exemplo: novo modelo de áudio
// Vision models
'llama-4-scout-17b-instruct': { rpm: 30, tpm: 6000 },
'llama-4-maverick-17b-instruct': { rpm: 30, tpm: 6000 },
// Text-to-Speech models
'playai-tts': { rpm: 50, tpm: 0 }, // Exemplo: novo modelo TTS
'playai-tts-arabic': { rpm: 50, tpm: 0 }, // Exemplo: novo modelo TTS
// Special endpoints (manter, se ainda relevantes para você)
'batch_processing': { rpm: 100, tpm: 0 },
'audio_whisper-large-v3': { rpm: 20, tpm: 0 },
'audio_whisper-large-v3-turbo': { rpm: 30, tpm: 0 },
'vision_llama-4-scout-17b-instruct': { rpm: 30, tpm: 6000 },
'vision_llama-4-maverick-17b-instruct': { rpm: 30, tpm: 6000 }
};
// Groq models configuration
export const GROQ_MODELS: Record<string, GroqModel> = Object.entries(modelsConfig.models).reduce((acc, [key, model]) => {
acc[key] = {
id: key,
name: model.name,
description: model.description,
capabilities: model.capabilities as ModelCapability[],
contextLength: model.contextLength,
maxTokens: model.maxTokens,
costPer1kTokens: model.costPer1kTokens,
speedRating: model.speedRating,
qualityRating: model.qualityRating,
rateLimits: {
requestsPerMinute: model.rateLimits.requestsPerMinute,
tokensPerMinute: model.rateLimits.tokensPerMinute
}
};
return acc;
}, {} as Record<string, GroqModel>);
// API endpoints
export const GROQ_API_BASE = '[https://api.groq.com/openai/v1](https://api.groq.com/openai/v1)';
export const API_ENDPOINTS = {
CHAT_COMPLETIONS: '/chat/completions',
AUDIO_TRANSCRIPTIONS: '/audio/transcriptions',
AUDIO_TRANSLATIONS: '/audio/translations',
BATCHES: '/batches',
FILES: '/files',
MODELS: '/models',
TEXT_TO_SPEECH: '/audio/speech' // Adicionado para TTS
} as const;
// Cache settings
export const CACHE_CONFIG = {
DEFAULT_TTL: 300, // 5 minutes
MAX_SIZE: 1000,
STRATEGY: 'lru' as const,
ENABLED: process.env.NODE_ENV !== 'test'
};
// Retry configuration
export const RETRY_CONFIG = {
MAX_RETRIES: 3,
INITIAL_DELAY: 1000, // 1 second
MAX_DELAY: 10000, // 10 seconds
BACKOFF_FACTOR: 2,
RETRYABLE_STATUS_CODES: [429, 500, 502, 503, 504],
RETRYABLE_ERROR_CODES: [
'rate_limit_exceeded',
'server_error',
'timeout',
'connection_error'
]
};
// Logging configuration
export const LOG_CONFIG = {
LEVEL: process.env.LOG_LEVEL || 'info',
MAX_FILE_SIZE: '5m',
MAX_FILES: 5,
DATE_PATTERN: 'YYYY-MM-DD',
STRUCTURED_LOGGING: true
};
// Health check configuration
export const HEALTH_CONFIG = {
CHECK_INTERVAL: 30000, // 30 seconds
TIMEOUT: 5000, // 5 seconds
CRITICAL_ERROR_THRESHOLD: 0.1, // 10% error rate
WARNING_ERROR_THRESHOLD: 0.05, // 5% error rate
CRITICAL_RESPONSE_TIME: 5000, // 5 seconds
WARNING_RESPONSE_TIME: 2000 // 2 seconds
};
// Metrics configuration
export const METRICS_CONFIG = {
COLLECTION_INTERVAL: 300000, // 5 minutes
RETENTION_DAYS: 30,
EXPORT_ENABLED: process.env.METRICS_EXPORT === 'true',
EXPORT_ENDPOINT: process.env.METRICS_ENDPOINT
};
// Security configuration
export const SECURITY_CONFIG = {
MAX_REQUEST_SIZE: '10mb',
ALLOWED_ORIGINS: process.env.ALLOWED_ORIGINS?.split(',') || ['*'],
API_KEY_HEADER: 'x-api-key',
RATE_LIMIT_WINDOW: 60000, // 1 minute
MAX_CONCURRENT_REQUESTS: 100
};
// File upload configuration
export const UPLOAD_CONFIG = {
MAX_FILE_SIZE: 25 * 1024 * 1024, // 25MB do limite do groq
ALLOWED_AUDIO_TYPES: [
'audio/mpeg',
'audio/wav',
'audio/flac',
'audio/m4a',
'audio/mp3',
'audio/mp4',
'audio/webm'
],
ALLOWED_IMAGE_TYPES: [
'image/jpeg',
'image/png',
'image/gif',
'image/webp'
],
TEMP_DIR: process.env.TEMP_DIR || '/tmp'
};
// Model selection weights
export const MODEL_SELECTION_WEIGHTS = {
SPEED: {
responseTime: 0.6,
throughput: 0.3,
availability: 0.1
},
QUALITY: {
accuracy: 0.5,
coherence: 0.3,
completeness: 0.2
},
COST: {
tokenCost: 0.7,
requestCost: 0.2,
efficiency: 0.1
}
};
// Error messages
export const ERROR_MESSAGES = {
INVALID_API_KEY: 'Invalid or missing Groq API key',
RATE_LIMIT_EXCEEDED: 'Rate limit exceeded. Please try again later.',
MODEL_NOT_FOUND: 'Specified model not found or not available',
INVALID_REQUEST: 'Invalid request parameters',
FILE_TOO_LARGE: 'File size exceeds maximum allowed limit',
UNSUPPORTED_FILE_TYPE: 'Unsupported file type',
NETWORK_ERROR: 'Network error occurred. Please try again.',
TIMEOUT_ERROR: 'Request timeout. Please try again.',
INTERNAL_ERROR: 'Internal server error occurred',
VALIDATION_ERROR: 'Request validation failed',
AUTHENTICATION_ERROR: 'Authentication failed',
AUTHORIZATION_ERROR: 'Authorization failed',
QUOTA_EXCEEDED: 'API quota exceeded',
SERVICE_UNAVAILABLE: 'Service temporarily unavailable'
};
// Success messages
export const SUCCESS_MESSAGES = {
TOOL_EXECUTED: 'Tool executed successfully',
FILE_UPLOADED: 'File uploaded successfully',
BATCH_CREATED: 'Batch job created successfully',
CACHE_HIT: 'Response served from cache',
MODEL_SELECTED: 'Optimal model selected automatically'
};
// Tool descriptions
export const TOOL_DESCRIPTIONS = {
TEXT_COMPLETION: 'Generate high-quality text completions using Groq\'s fastest LLM models with intelligent routing based on complexity and requirements.',
AUDIO_TRANSCRIPTION: 'Transcribe audio files to text using Groq\'s Whisper models with support for multiple languages and output formats.',
VISION_ANALYSIS: 'Analyze images and extract insights using Groq\'s multimodal vision models with customizable analysis types.',
BATCH_PROCESSING: 'Process large batches of requests efficiently with 25% cost savings and automatic job management.',
TEXT_TO_SPEECH: 'Synthesize high-quality audio from text using Groq\'s Text-to-Speech models with various voice options.' // Adicionado para TTS
};
// Environment validation
export const REQUIRED_ENV_VARS = [
'GROQ_API_KEY'
];
export const OPTIONAL_ENV_VARS = [
'LOG_LEVEL',
'NODE_ENV',
'CACHE_ENABLED',
'METRICS_ENABLED',
'TEMP_DIR',
'ALLOWED_ORIGINS',
'METRICS_ENDPOINT'
];
// Version and build info
export const VERSION_INFO = {
VERSION: '1.0.0',
BUILD_DATE: new Date().toISOString(),
NODE_VERSION: process.version,
PLATFORM: process.platform,
ARCH: process.arch
};
// Default configurations for different environments
export const ENV_CONFIGS = {
development: {
logLevel: 'debug',
cacheEnabled: true,
metricsEnabled: true,
rateLimitEnabled: false,
corsEnabled: true
},
production: {
logLevel: 'info',
cacheEnabled: true,
metricsEnabled: true,
rateLimitEnabled: true,
corsEnabled: false
},
test: {
logLevel: 'error',
cacheEnabled: false,
metricsEnabled: false,
rateLimitEnabled: false,
corsEnabled: true
}
};
// Get current environment configuration
export const getCurrentEnvConfig = () => {
const env = process.env.NODE_ENV || 'development';
return ENV_CONFIGS[env as keyof typeof ENV_CONFIGS] || ENV_CONFIGS.development;
};
