isa-model 0.3.0__tar.gz → 0.3.2__tar.gz

{isa_model-0.3.0 → isa_model-0.3.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: isa_model
-Version: 0.3.0
+Version: 0.3.2
 Summary: Unified AI model serving framework
 Author: isA_Model Contributors
 Classifier: Development Status :: 3 - Alpha

{isa_model-0.3.0 → isa_model-0.3.2}/isa_model/core/model_manager.py

@@ -2,7 +2,7 @@ from typing import Dict, Optional, List, Any
 import logging
 from pathlib import Path
 from huggingface_hub import hf_hub_download, snapshot_download
-from huggingface_hub.utils import HfHubHTTPError
+from huggingface_hub.errors import HfHubHTTPError
 from .model_storage import ModelStorage, LocalModelStorage
 from .model_registry import ModelRegistry, ModelType, ModelCapability
 
@@ -11,19 +11,81 @@ logger = logging.getLogger(__name__)
 class ModelManager:
     """Model management service for handling model downloads, versions, and caching"""
     
+    # 统一的模型计费信息 (per 1M tokens)
+    MODEL_PRICING = {
+        # OpenAI Models
+        "openai": {
+            "gpt-4o-mini": {"input": 0.15, "output": 0.6},
+            "gpt-4.1-mini": {"input": 0.4, "output": 1.6},
+            "gpt-4.1-nano": {"input": 0.1, "output": 0.4},
+            "gpt-4o": {"input": 5.0, "output": 15.0},
+            "gpt-4-turbo": {"input": 10.0, "output": 30.0},
+            "gpt-4": {"input": 30.0, "output": 60.0},
+            "gpt-3.5-turbo": {"input": 0.5, "output": 1.5},
+            "text-embedding-3-small": {"input": 0.02, "output": 0.0},
+            "text-embedding-3-large": {"input": 0.13, "output": 0.0},
+            "whisper-1": {"input": 6.0, "output": 0.0},
+            "tts-1": {"input": 15.0, "output": 0.0},
+            "tts-1-hd": {"input": 30.0, "output": 0.0},
+        },
+        # Ollama Models (免费本地模型)
+        "ollama": {
+            "llama3.2:3b-instruct-fp16": {"input": 0.0, "output": 0.0},
+            "llama3.2-vision:latest": {"input": 0.0, "output": 0.0},
+            "bge-m3": {"input": 0.0, "output": 0.0},
+        },
+        # Replicate Models
+        "replicate": {
+            "black-forest-labs/flux-schnell": {"input": 3.0, "output": 0.0},  # $3 per 1000 images
+            "black-forest-labs/flux-kontext-pro": {"input": 40.0, "output": 0.0},  # $0.04 per image = $40 per 1000 images
+            "meta/meta-llama-3-8b-instruct": {"input": 0.05, "output": 0.25},
+            "kokoro-82m": {"input": 0.0, "output": 0.4},  # ~$0.0004 per second
+            "jaaari/kokoro-82m:f559560eb822dc509045f3921a1921234918b91739db4bf3daab2169b71c7a13": {"input": 0.0, "output": 0.4},
+        }
+    }
+    
     def __init__(self, 
                  storage: Optional[ModelStorage] = None,
                  registry: Optional[ModelRegistry] = None):
         self.storage = storage or LocalModelStorage()
         self.registry = registry or ModelRegistry()
     
+    def get_model_pricing(self, provider: str, model_name: str) -> Dict[str, float]:
+        """获取模型定价信息"""
+        return self.MODEL_PRICING.get(provider, {}).get(model_name, {"input": 0.0, "output": 0.0})
+    
+    def calculate_cost(self, provider: str, model_name: str, input_tokens: int, output_tokens: int) -> float:
+        """计算请求成本"""
+        pricing = self.get_model_pricing(provider, model_name)
+        input_cost = (input_tokens / 1_000_000) * pricing["input"]
+        output_cost = (output_tokens / 1_000_000) * pricing["output"]
+        return input_cost + output_cost
+    
+    def get_cheapest_model(self, provider: str, model_type: str = "llm") -> Optional[str]:
+        """获取最便宜的模型"""
+        provider_models = self.MODEL_PRICING.get(provider, {})
+        if not provider_models:
+            return None
+        
+        # 计算每个模型的平均成本 (假设输入输出比例 1:1)
+        cheapest_model = None
+        lowest_cost = float('inf')
+        
+        for model_name, pricing in provider_models.items():
+            avg_cost = (pricing["input"] + pricing["output"]) / 2
+            if avg_cost < lowest_cost:
+                lowest_cost = avg_cost
+                cheapest_model = model_name
+        
+        return cheapest_model
+
     async def get_model(self, 
                        model_id: str, 
                        repo_id: str,
                        model_type: ModelType,
                        capabilities: List[ModelCapability],
                        revision: Optional[str] = None,
-                       force_download: bool = False) -> Path:
+                       force_download: bool = False) -> Optional[Path]:
         """
         Get model files, downloading if necessary
         
@@ -36,7 +98,7 @@ class ModelManager:
             force_download: Force re-download even if cached
             
         Returns:
-            Path to the model files
+            Path to the model files or None if failed
         """
         # Check if model is already downloaded
         if not force_download:
@@ -80,7 +142,10 @@ class ModelManager:
             
         except HfHubHTTPError as e:
             logger.error(f"Failed to download model {model_id}: {e}")
-            raise
+            return None
+        except Exception as e:
+            logger.error(f"Unexpected error downloading model {model_id}: {e}")
+            return None
     
     async def list_models(self) -> List[Dict[str, Any]]:
         """List all downloaded models with their metadata"""

isa_model-0.3.2/isa_model/inference/ai_factory.py

@@ -0,0 +1,550 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""
+Simplified AI Factory for creating inference services
+Uses the new service architecture with proper base classes and centralized API key management
+"""
+
+from typing import Dict, Type, Any, Optional, Tuple, List, TYPE_CHECKING, cast
+import logging
+from isa_model.inference.providers.base_provider import BaseProvider
+from isa_model.inference.services.base_service import BaseService
+from isa_model.inference.base import ModelType
+from isa_model.inference.services.vision.base_vision_service import BaseVisionService
+from isa_model.inference.services.vision.base_image_gen_service import BaseImageGenService
+
+if TYPE_CHECKING:
+    from isa_model.inference.services.audio.base_stt_service import BaseSTTService
+    from isa_model.inference.services.audio.base_tts_service import BaseTTSService
+
+logger = logging.getLogger(__name__)
+
+class AIFactory:
+    """
+    Simplified Factory for creating AI services with proper inheritance hierarchy
+    API key management is handled by individual providers
+    """
+    
+    _instance = None
+    _is_initialized = False
+    
+    def __new__(cls):
+        if cls._instance is None:
+            cls._instance = super().__new__(cls)
+        return cls._instance
+    
+    def __init__(self):
+        """Initialize the AI Factory."""
+        if not self._is_initialized:
+            self._providers: Dict[str, Type[BaseProvider]] = {}
+            self._services: Dict[Tuple[str, ModelType], Type[BaseService]] = {}
+            self._cached_services: Dict[str, BaseService] = {}
+            self._initialize_services()
+            AIFactory._is_initialized = True
+    
+    def _initialize_services(self):
+        """Initialize available providers and services"""
+        try:
+            # Register Ollama services
+            self._register_ollama_services()
+            
+            # Register OpenAI services
+            self._register_openai_services()
+            
+            # Register Replicate services
+            self._register_replicate_services()
+            
+            logger.info("AI Factory initialized with centralized provider API key management")
+            
+        except Exception as e:
+            logger.error(f"Error initializing services: {e}")
+            logger.warning("Some services may not be available")
+    
+    def _register_ollama_services(self):
+        """Register Ollama provider and services"""
+        try:
+            from isa_model.inference.providers.ollama_provider import OllamaProvider
+            from isa_model.inference.services.llm.ollama_llm_service import OllamaLLMService
+            from isa_model.inference.services.embedding.ollama_embed_service import OllamaEmbedService
+            from isa_model.inference.services.vision.ollama_vision_service import OllamaVisionService
+            
+            self.register_provider('ollama', OllamaProvider)
+            self.register_service('ollama', ModelType.LLM, OllamaLLMService)
+            self.register_service('ollama', ModelType.EMBEDDING, OllamaEmbedService)
+            self.register_service('ollama', ModelType.VISION, OllamaVisionService)
+            
+            logger.info("Ollama services registered successfully")
+            
+        except ImportError as e:
+            logger.warning(f"Ollama services not available: {e}")
+    
+    def _register_openai_services(self):
+        """Register OpenAI provider and services"""
+        try:
+            from isa_model.inference.providers.openai_provider import OpenAIProvider
+            from isa_model.inference.services.llm.openai_llm_service import OpenAILLMService
+            from isa_model.inference.services.audio.openai_tts_service import OpenAITTSService
+            from isa_model.inference.services.audio.openai_stt_service import OpenAISTTService
+            from isa_model.inference.services.embedding.openai_embed_service import OpenAIEmbedService
+            from isa_model.inference.services.vision.openai_vision_service import OpenAIVisionService
+            
+            self.register_provider('openai', OpenAIProvider)
+            self.register_service('openai', ModelType.LLM, OpenAILLMService)
+            self.register_service('openai', ModelType.AUDIO, OpenAITTSService)
+            self.register_service('openai', ModelType.EMBEDDING, OpenAIEmbedService)
+            self.register_service('openai', ModelType.VISION, OpenAIVisionService)
+            
+            logger.info("OpenAI services registered successfully")
+            
+        except ImportError as e:
+            logger.warning(f"OpenAI services not available: {e}")
+    
+    def _register_replicate_services(self):
+        """Register Replicate provider and services"""
+        try:
+            from isa_model.inference.providers.replicate_provider import ReplicateProvider
+            from isa_model.inference.services.vision.replicate_image_gen_service import ReplicateImageGenService
+            from isa_model.inference.services.audio.replicate_tts_service import ReplicateTTSService
+            
+            self.register_provider('replicate', ReplicateProvider)
+            self.register_service('replicate', ModelType.VISION, ReplicateImageGenService)
+            self.register_service('replicate', ModelType.AUDIO, ReplicateTTSService)
+            
+            logger.info("Replicate services registered successfully")
+            
+        except ImportError as e:
+            logger.warning(f"Replicate services not available: {e}")
+    
+    def register_provider(self, name: str, provider_class: Type[BaseProvider]) -> None:
+        """Register an AI provider"""
+        self._providers[name] = provider_class
+    
+    def register_service(self, provider_name: str, model_type: ModelType, 
+                        service_class: Type[BaseService]) -> None:
+        """Register a service type with its provider"""
+        self._services[(provider_name, model_type)] = service_class
+    
+    def create_service(self, provider_name: str, model_type: ModelType, 
+                      model_name: str, config: Optional[Dict[str, Any]] = None) -> BaseService:
+        """Create a service instance with provider-managed configuration"""
+        try:
+            cache_key = f"{provider_name}_{model_type}_{model_name}"
+            
+            if cache_key in self._cached_services:
+                return self._cached_services[cache_key]
+            
+            # Get provider and service classes
+            provider_class = self._providers.get(provider_name)
+            service_class = self._services.get((provider_name, model_type))
+            
+            if not provider_class:
+                raise ValueError(f"No provider registered for '{provider_name}'")
+            
+            if not service_class:
+                raise ValueError(
+                    f"No service registered for provider '{provider_name}' and model type '{model_type}'"
+                )
+            
+            # Create provider with user config (provider handles .env loading)
+            provider = provider_class(config=config)
+            service = service_class(provider=provider, model_name=model_name)
+            
+            self._cached_services[cache_key] = service
+            return service
+            
+        except Exception as e:
+            logger.error(f"Error creating service: {e}")
+            raise
+    
+    # Convenient methods for common services with updated defaults
+    def get_llm_service(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                       config: Optional[Dict[str, Any]] = None) -> BaseService:
+        """
+        Get a LLM service instance with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: OpenAI="gpt-4.1-nano", Ollama="llama3.2:3b")
+            provider: Provider name (defaults to 'openai' for production, 'ollama' for dev)
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+                   Can include: streaming=True/False, temperature, max_tokens, etc.
+            
+        Returns:
+            LLM service instance
+        """
+        # Set defaults based on provider
+        if provider == "openai":
+            final_model_name = model_name or "gpt-4.1-nano"
+            final_provider = provider
+        elif provider == "ollama":
+            final_model_name = model_name or "llama3.2:3b-instruct-fp16"
+            final_provider = provider
+        else:
+            # Default provider selection - OpenAI with cheapest model
+            final_provider = provider or "openai"
+            if final_provider == "openai":
+                final_model_name = model_name or "gpt-4.1-nano"
+            else:
+                final_model_name = model_name or "llama3.2:3b-instruct-fp16"
+        
+        return self.create_service(final_provider, ModelType.LLM, final_model_name, config)
+    
+    def get_embedding_service(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                             config: Optional[Dict[str, Any]] = None) -> BaseService:
+        """
+        Get an embedding service instance with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: OpenAI="text-embedding-3-small", Ollama="bge-m3")
+            provider: Provider name (defaults to 'openai' for production, 'ollama' for dev)
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            Embedding service instance
+        """
+        # Set defaults based on provider
+        if provider == "openai":
+            final_model_name = model_name or "text-embedding-3-small"
+            final_provider = provider
+        elif provider == "ollama":
+            final_model_name = model_name or "bge-m3"
+            final_provider = provider
+        else:
+            # Default provider selection
+            final_provider = provider or "openai"
+            if final_provider == "openai":
+                final_model_name = model_name or "text-embedding-3-small"
+            else:
+                final_model_name = model_name or "bge-m3"
+        
+        return self.create_service(final_provider, ModelType.EMBEDDING, final_model_name, config)
+    
+    def get_vision_service(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                          config: Optional[Dict[str, Any]] = None) -> BaseVisionService:
+        """
+        Get a vision service instance with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: OpenAI="gpt-4.1-mini", Ollama="gemma3:4b")
+            provider: Provider name (defaults to 'openai' for production, 'ollama' for dev)
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            Vision service instance
+        """
+        # Set defaults based on provider
+        if provider == "openai":
+            final_model_name = model_name or "gpt-4.1-mini"
+            final_provider = provider
+        elif provider == "ollama":
+            final_model_name = model_name or "llama3.2-vision:latest"
+            final_provider = provider
+        else:
+            # Default provider selection
+            final_provider = provider or "openai"
+            if final_provider == "openai":
+                final_model_name = model_name or "gpt-4.1-mini"
+            else:
+                final_model_name = model_name or "llama3.2-vision:latest"
+        
+        return cast(BaseVisionService, self.create_service(final_provider, ModelType.VISION, final_model_name, config))
+    
+    def get_image_generation_service(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                                   config: Optional[Dict[str, Any]] = None) -> 'BaseImageGenService':
+        """
+        Get an image generation service instance with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: "black-forest-labs/flux-schnell" for production)
+            provider: Provider name (defaults to 'replicate')
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            Image generation service instance
+        """
+        # Set defaults based on provider
+        final_provider = provider or "replicate"
+        if final_provider == "replicate":
+            final_model_name = model_name or "black-forest-labs/flux-schnell"
+        else:
+            final_model_name = model_name or "black-forest-labs/flux-schnell"
+        
+        return cast('BaseImageGenService', self.create_service(final_provider, ModelType.VISION, final_model_name, config))
+    
+    def get_img(self, type: str = "t2i", model_name: Optional[str] = None, provider: Optional[str] = None,
+                config: Optional[Dict[str, Any]] = None) -> 'BaseImageGenService':
+        """
+        Get an image generation service with type-specific defaults
+        
+        Args:
+            type: Image generation type:
+                  - "t2i" (text-to-image): Uses flux-schnell ($3 per 1000 images)
+                  - "i2i" (image-to-image): Uses flux-kontext-pro ($0.04 per image)
+            model_name: Optional model name override
+            provider: Provider name (defaults to 'replicate')
+            config: Optional configuration dictionary
+            
+        Returns:
+            Image generation service instance
+            
+        Usage:
+            # Text-to-image (default)
+            img_service = AIFactory().get_img()
+            img_service = AIFactory().get_img(type="t2i")
+            
+            # Image-to-image
+            img_service = AIFactory().get_img(type="i2i")
+            
+            # Custom model
+            img_service = AIFactory().get_img(type="t2i", model_name="custom-model")
+        """
+        # Set defaults based on type
+        final_provider = provider or "replicate"
+        
+        if type == "t2i":
+            # Text-to-image: flux-schnell
+            final_model_name = model_name or "black-forest-labs/flux-schnell"
+        elif type == "i2i":
+            # Image-to-image: flux-kontext-pro
+            final_model_name = model_name or "black-forest-labs/flux-kontext-pro"
+        else:
+            raise ValueError(f"Unknown image generation type: {type}. Use 't2i' or 'i2i'")
+        
+        return cast('BaseImageGenService', self.create_service(final_provider, ModelType.VISION, final_model_name, config))
+    
+    def get_audio_service(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                         config: Optional[Dict[str, Any]] = None) -> BaseService:
+        """
+        Get an audio service instance (TTS) with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: OpenAI="tts-1")
+            provider: Provider name (defaults to 'openai')
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            Audio service instance
+        """
+        # Set defaults based on provider
+        final_provider = provider or "openai"
+        if final_provider == "openai":
+            final_model_name = model_name or "tts-1"
+        else:
+            final_model_name = model_name or "tts-1"
+        
+        return self.create_service(final_provider, ModelType.AUDIO, final_model_name, config)
+    
+    def get_tts_service(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                       config: Optional[Dict[str, Any]] = None) -> 'BaseTTSService':
+        """
+        Get a Text-to-Speech service instance with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: Replicate="kokoro-82m", OpenAI="tts-1")
+            provider: Provider name (defaults to 'replicate' for production, 'openai' for dev)
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            TTS service instance
+        """
+        # Set defaults based on provider
+        if provider == "replicate":
+            model_name = model_name or "kokoro-82m"
+        elif provider == "openai":
+            model_name = model_name or "tts-1"
+        else:
+            # Default provider selection
+            provider = provider or "replicate"
+            if provider == "replicate":
+                model_name = model_name or "kokoro-82m"
+            else:
+                model_name = model_name or "tts-1"
+        
+        # Ensure model_name is never None
+        if model_name is None:
+            model_name = "tts-1"
+        
+        if provider == "replicate":
+            from isa_model.inference.services.audio.replicate_tts_service import ReplicateTTSService
+            from isa_model.inference.providers.replicate_provider import ReplicateProvider
+            
+            # Use full model name for Replicate
+            if model_name == "kokoro-82m":
+                model_name = "jaaari/kokoro-82m:f559560eb822dc509045f3921a1921234918b91739db4bf3daab2169b71c7a13"
+            
+            provider_instance = ReplicateProvider(config=config)
+            return ReplicateTTSService(provider=provider_instance, model_name=model_name)
+        else:
+            return cast('BaseTTSService', self.get_audio_service(model_name, provider, config))
+    
+    def get_stt_service(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                       config: Optional[Dict[str, Any]] = None) -> 'BaseSTTService':
+        """
+        Get a Speech-to-Text service instance with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: "whisper-1")
+            provider: Provider name (defaults to 'openai')
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            STT service instance
+        """
+        # Set defaults based on provider
+        provider = provider or "openai"
+        if provider == "openai":
+            model_name = model_name or "whisper-1"
+        
+        # Ensure model_name is never None
+        if model_name is None:
+            model_name = "whisper-1"
+        
+        from isa_model.inference.services.audio.openai_stt_service import OpenAISTTService
+        from isa_model.inference.providers.openai_provider import OpenAIProvider
+        
+        # Create provider and service directly with config
+        provider_instance = OpenAIProvider(config=config)
+        return OpenAISTTService(provider=provider_instance, model_name=model_name)
+    
+    def get_available_services(self) -> Dict[str, List[str]]:
+        """Get information about available services"""
+        services = {}
+        for (provider, model_type), service_class in self._services.items():
+            if provider not in services:
+                services[provider] = []
+            services[provider].append(f"{model_type.value}: {service_class.__name__}")
+        return services
+    
+    def clear_cache(self):
+        """Clear the service cache"""
+        self._cached_services.clear()
+        logger.info("Service cache cleared")
+    
+    @classmethod
+    def get_instance(cls) -> 'AIFactory':
+        """Get the singleton instance"""
+        if cls._instance is None:
+            cls._instance = cls()
+        return cls._instance
+    
+    # Alias method for cleaner API
+    def get_llm(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                config: Optional[Dict[str, Any]] = None) -> BaseService:
+        """
+        Alias for get_llm_service with cleaner naming
+        
+        Usage:
+            llm = AIFactory().get_llm()  # Uses gpt-4.1-nano by default
+            llm = AIFactory().get_llm(model_name="llama3.2", provider="ollama")
+            llm = AIFactory().get_llm(model_name="gpt-4.1-mini", provider="openai", config={"streaming": True})
+        """
+        return self.get_llm_service(model_name, provider, config)
+    
+    def get_embed(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                     config: Optional[Dict[str, Any]] = None) -> 'BaseEmbedService':
+        """
+        Get embedding service with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: OpenAI="text-embedding-3-small", Ollama="bge-m3")
+            provider: Provider name (defaults to 'openai' for production)
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            Embedding service instance
+            
+        Usage:
+            # Default (OpenAI text-embedding-3-small)
+            embed = AIFactory().get_embed()
+            
+            # Custom model
+            embed = AIFactory().get_embed(model_name="text-embedding-3-large", provider="openai")
+            
+            # Development (Ollama)
+            embed = AIFactory().get_embed(provider="ollama")
+        """
+        return self.get_embedding_service(model_name, provider, config)
+
+    def get_stt(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                config: Optional[Dict[str, Any]] = None) -> 'BaseSTTService':
+        """
+        Get Speech-to-Text service with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: "whisper-1")
+            provider: Provider name (defaults to 'openai')
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            STT service instance
+            
+        Usage:
+            # Default (OpenAI whisper-1)
+            stt = AIFactory().get_stt()
+            
+            # Custom configuration
+            stt = AIFactory().get_stt(model_name="whisper-1", provider="openai")
+        """
+        return self.get_stt_service(model_name, provider, config)
+
+    def get_tts(self, model_name: Optional[str] = None, provider: Optional[str] = None,
+                config: Optional[Dict[str, Any]] = None) -> 'BaseTTSService':
+        """
+        Get Text-to-Speech service with automatic defaults
+        
+        Args:
+            model_name: Name of the model to use (defaults: Replicate="kokoro-82m", OpenAI="tts-1")
+            provider: Provider name (defaults to 'replicate' for production, 'openai' for dev)
+            config: Optional configuration dictionary (auto-loads from .env if not provided)
+            
+        Returns:
+            TTS service instance
+            
+        Usage:
+            # Default (Replicate kokoro-82m)
+            tts = AIFactory().get_tts()
+            
+            # Development (OpenAI tts-1)
+            tts = AIFactory().get_tts(provider="openai")
+            
+            # Custom model
+            tts = AIFactory().get_tts(model_name="tts-1-hd", provider="openai")
+        """
+        return self.get_tts_service(model_name, provider, config)
+    
+    def get_vision_model(self, model_name: str, provider: str,
+                        config: Optional[Dict[str, Any]] = None) -> BaseService:
+        """Alias for get_vision_service and get_image_generation_service"""
+        if provider == "replicate":
+            return self.get_image_generation_service(model_name, provider, config)
+        else:
+            return self.get_vision_service(model_name, provider, config)
+    
+    def get_vision(
+        self,
+        model_name: Optional[str] = None,
+        provider: Optional[str] = None,
+        config: Optional[Dict[str, Any]] = None
+    ) -> 'BaseVisionService':
+        """
+        Get vision service with automatic defaults
+        
+        Args:
+            model_name: Model name (default: gpt-4.1-nano)
+            provider: Provider name (default: openai)
+            config: Optional configuration override
+            
+        Returns:
+            Vision service instance
+        """
+        # Set defaults
+        if provider is None:
+            provider = "openai"
+        if model_name is None:
+            model_name = "gpt-4.1-nano"
+        
+        return self.get_vision_service(
+            model_name=model_name,
+            provider=provider,
+            config=config
+        )