feat(model): add OpenAI provider

.env.example

@@ -50,7 +50,7 @@ GOOGLE_API_KEY=
 # If you obtained your key from https://siliconflow.com, you need to update siliconflow.yaml manually.
 SILICONFLOW_API_KEY=
 
-# Get your API key from: https://platform.openai.com/api-keys TODO: will support in next PR. 
+# Get your API key from: https://platform.openai.com/api-keys
 OPENAI_API_KEY=
 
 

python/configs/config.yaml

@@ -30,6 +30,10 @@ models:
     google:
       config_file: "providers/google.yaml"
       api_key_env: "GOOGLE_API_KEY"
+
+    openai:
+      config_file: "providers/openai.yaml"
+      api_key_env: "OPENAI_API_KEY"
 
 # Agent Configuration
 agents:

python/configs/providers/openai.yaml

@@ -0,0 +1,69 @@
+# ============================================
+# OpenAI Provider Configuration
+# ============================================
+name: "OpenAI"
+provider_type: "openai"
+
+enabled: true # Default is true if not specified
+
+# Connection Configuration
+connection:
+  base_url: "https://api.openai.com/v1"
+  api_key_env: "OPENAI_API_KEY"
+
+# Default model if none specified
+default_model: "gpt-5-2025-08-07"
+
+# Available Models
+models:
+
+  - id: "gpt-5-2025-08-07"
+    name: "GPT-5 2025-08-07"
+    context_length: 400000
+    max_output_tokens: 128000
+    description: "Most capable GPT-5 model with 400K context window"
+    supported_inputs: ["text"]
+    supported_outputs: ["text"]
+
+  - id: "gpt-5-mini-2025-08-07"
+    name: "GPT-5 Mini 2025-08-07"
+    context_length: 400000
+    max_output_tokens: 128000
+    description: "Lightweight GPT-5 model optimized for efficiency"
+    supported_inputs: ["text"]
+    supported_outputs: ["text"]
+
+  - id: "gpt-4.1-2025-04-14"
+    name: "GPT-4.1 2025-04-14"
+    context_length: 1047576
+    max_output_tokens: 32768
+    description: "Most capable GPT-4.1 model with 1M context window (without organization verification)"
+    supported_inputs: ["text"]
+    supported_outputs: ["text"]
+
+# ============================================
+# Embedding Models Configuration
+# ============================================
+embedding:
+  # Default embedding model
+  default_model: "text-embedding-3-small"
+
+  # Default parameters
+  defaults:
+    encoding_format: "float"
+
+  # Available embedding models
+  models:
+    - id: "text-embedding-3-small"
+      name: "Text Embedding 3 Small"
+      dimensions: 1536
+      max_input: 8192
+      performance_mteb: 62.3
+      description: "Small embedding model with good performance and efficiency"
+
+    - id: "text-embedding-3-large"
+      name: "Text Embedding 3 Large"
+      dimensions: 3072
+      max_input: 8192
+      performance_mteb: 64.6
+      description: "Large embedding model with higher performance"

python/valuecell/adapters/models/factory.py

@@ -252,6 +252,63 @@ def create_embedder(self, model_id: Optional[str] = None, **kwargs):
         )
 
 
+class OpenAIProvider(ModelProvider):
+    """OpenAI model provider"""
+
+    def create_model(self, model_id: Optional[str] = None, **kwargs):
+        """Create OpenAI model via agno"""
+        try:
+            from agno.models.openai import OpenAIChat
+        except ImportError:
+            raise ImportError(
+                "agno package not installed. Install with: pip install agno"
+            )
+
+        model_id = model_id or self.config.default_model
+        params = {**self.config.parameters, **kwargs}
+
+        logger.info(f"Creating OpenAI model: {model_id}")
+
+        return OpenAIChat(
+            id=model_id,
+            api_key=self.config.api_key,
+            temperature=params.get("temperature"),
+            max_tokens=params.get("max_tokens"),
+            top_p=params.get("top_p"),
+            frequency_penalty=params.get("frequency_penalty"),
+            presence_penalty=params.get("presence_penalty"),
+        )
+
+    def create_embedder(self, model_id: Optional[str] = None, **kwargs):
+        """Create embedder via OpenAI"""
+        try:
+            from agno.knowledge.embedder.openai import OpenAIEmbedder
+        except ImportError:
+            raise ImportError("agno package not installed")
+
+        # Use provided model_id or default embedding model
+        model_id = model_id or self.config.default_embedding_model
+
+        if not model_id:
+            raise ValueError(
+                f"No embedding model specified for provider '{self.config.name}'"
+            )
+
+        # Merge parameters: provider embedding defaults < kwargs
+        params = {**self.config.embedding_parameters, **kwargs}
+
+        logger.info(f"Creating OpenAI embedder: {model_id}")
+
+        return OpenAIEmbedder(
+            id=model_id,
+            api_key=self.config.api_key,
+            dimensions=int(params.get("dimensions", 1536))
+            if params.get("dimensions")
+            else None,
+            encoding_format=params.get("encoding_format", "float"),
+        )
+
+
 class ModelFactory:
     """
     Factory for creating model instances with provider abstraction
@@ -269,6 +326,7 @@ class ModelFactory:
         "google": GoogleProvider,
         "azure": AzureProvider,
         "siliconflow": SiliconFlowProvider,
+        "openai": OpenAIProvider,
     }
 
     def __init__(self, config_manager: Optional[ConfigManager] = None):

python/valuecell/config/manager.py

@@ -132,6 +132,7 @@ def primary_provider(self) -> str:
                     "openrouter",
                     "siliconflow",
                     "google",
+                    "openai",
                 ]
 
                 for preferred in preferred_order:

python/valuecell/utils/model.py

@@ -15,6 +15,7 @@
 
 from agno.models.base import Model as AgnoModel
 from agno.models.google import Gemini as AgnoGeminiModel
+from agno.models.openai import OpenAIChat as AgnoOpenAIChatModel
 
 from valuecell.adapters.models.factory import (
     create_embedder,
@@ -32,6 +33,11 @@ def model_should_use_json_mode(model: AgnoModel) -> bool:
         name = getattr(model, "name", None)
         if provider == AgnoGeminiModel.provider and name == AgnoGeminiModel.name:
             return True
+        if (
+            provider == AgnoOpenAIChatModel.provider
+            and name == AgnoOpenAIChatModel.name
+        ):
+            return True
     except Exception:
         # Any unexpected condition falls back to standard (non-JSON) mode
         return False