Implement Small Model in MCP Server; Default to 4.1-nano (#516)

Add small LLM model support and update README

- Introduced a new small LLM model configuration option (`SMALL_LLM_MODEL`) in `GraphitiLLMConfig`.
- Updated environment variable documentation to include `SMALL_MODEL_NAME`.
- Added command-line argument support for `--small-model` to override the small model name.
- Enhanced logging for small model configuration to improve user feedback.

These changes enhance flexibility in model selection for LLM operations.

mcp_server/README.md

@@ -75,7 +75,9 @@ The server uses the following environment variables:
 - `NEO4J_PASSWORD`: Neo4j password (default: `demodemo`)
 - `OPENAI_API_KEY`: OpenAI API key (required for LLM operations)
 - `OPENAI_BASE_URL`: Optional base URL for OpenAI API
-- `MODEL_NAME`: Optional model name to use for LLM inference
+- `MODEL_NAME`: OpenAI model name to use for LLM operations.
+- `SMALL_MODEL_NAME`: OpenAI model name to use for smaller LLM operations.
+- `LLM_TEMPERATURE`: Temperature for LLM responses (0.0-2.0).
 - `AZURE_OPENAI_ENDPOINT`: Optional Azure OpenAI endpoint URL
 - `AZURE_OPENAI_DEPLOYMENT_NAME`: Optional Azure OpenAI deployment name
 - `AZURE_OPENAI_API_VERSION`: Optional Azure OpenAI API version
@@ -101,10 +103,12 @@ uv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse
 
 Available arguments:
 
-- `--model`: Specify the model name to use with the LLM client
+- `--model`: Overrides the `MODEL_NAME` environment variable.
+- `--small-model`: Overrides the `SMALL_MODEL_NAME` environment variable.
+- `--temperature`: Overrides the `LLM_TEMPERATURE` environment variable.
 - `--transport`: Choose the transport method (sse or stdio, default: sse)
 - `--group-id`: Set a namespace for the graph (optional). If not provided, defaults to "default".
-- `--destroy-graph`: Destroy all Graphiti graphs (use with caution)
+- `--destroy-graph`: If set, destroys all Graphiti graphs on startup.
 - `--use-custom-entities`: Enable entity extraction using the predefined ENTITY_TYPES
 
 ### Docker Deployment

mcp_server/graphiti_mcp_server.py

@@ -38,6 +38,7 @@ from graphiti_core.utils.maintenance.graph_data_operations import clear_data
 load_dotenv()
 
 DEFAULT_LLM_MODEL = 'gpt-4.1-mini'
+SMALL_LLM_MODEL = 'gpt-4.1-nano'
 DEFAULT_EMBEDDER_MODEL = 'text-embedding-3-small'
 
 
@@ -187,6 +188,7 @@ class GraphitiLLMConfig(BaseModel):
 
     api_key: str | None = None
     model: str = DEFAULT_LLM_MODEL
+    small_model: str = SMALL_LLM_MODEL
     temperature: float = 0.0
     azure_openai_endpoint: str | None = None
     azure_openai_deployment_name: str | None = None
@@ -200,6 +202,10 @@ class GraphitiLLMConfig(BaseModel):
         model_env = os.environ.get('MODEL_NAME', '')
         model = model_env if model_env.strip() else DEFAULT_LLM_MODEL
 
+        # Get small_model from environment, or use default if not set or empty
+        small_model_env = os.environ.get('SMALL_MODEL_NAME', '')
+        small_model = small_model_env if small_model_env.strip() else SMALL_LLM_MODEL
+
         azure_openai_endpoint = os.environ.get('AZURE_OPENAI_ENDPOINT', None)
         azure_openai_api_version = os.environ.get('AZURE_OPENAI_API_VERSION', None)
         azure_openai_deployment_name = os.environ.get('AZURE_OPENAI_DEPLOYMENT_NAME', None)
@@ -222,6 +228,7 @@ class GraphitiLLMConfig(BaseModel):
             return cls(
                 api_key=os.environ.get('OPENAI_API_KEY'),
                 model=model,
+                small_model=small_model,
                 temperature=float(os.environ.get('LLM_TEMPERATURE', '0.0')),
             )
         else:
@@ -244,6 +251,8 @@ class GraphitiLLMConfig(BaseModel):
                 api_key=api_key,
                 azure_openai_api_version=azure_openai_api_version,
                 azure_openai_deployment_name=azure_openai_deployment_name,
+                model=model,
+                small_model=small_model,
                 temperature=float(os.environ.get('LLM_TEMPERATURE', '0.0')),
             )
 
@@ -262,6 +271,12 @@ class GraphitiLLMConfig(BaseModel):
                 # Log that empty model was provided and default is used
                 logger.warning(f'Empty model name provided, using default: {DEFAULT_LLM_MODEL}')
 
+        if hasattr(args, 'small_model') and args.small_model:
+            if args.small_model.strip():
+                config.small_model = args.small_model
+            else:
+                logger.warning(f'Empty small_model name provided, using default: {SMALL_LLM_MODEL}')
+
         if hasattr(args, 'temperature') and args.temperature is not None:
             config.temperature = args.temperature
 
@@ -301,7 +316,7 @@ class GraphitiLLMConfig(BaseModel):
             return None
 
         llm_client_config = LLMConfig(
-            api_key=self.api_key, model=self.model, small_model=self.model
+            api_key=self.api_key, model=self.model, small_model=self.small_model
         )
 
         # Set temperature
@@ -315,7 +330,9 @@ class GraphitiLLMConfig(BaseModel):
             client = self.create_client()
             return OpenAIRerankerClient(client=client)
         else:
-            llm_client_config = LLMConfig(api_key=self.api_key, model=self.model)
+            llm_client_config = LLMConfig(
+                api_key=self.api_key, model=self.model, small_model=self.small_model
+            )
             return OpenAIRerankerClient(config=llm_client_config)
 
 
@@ -1134,6 +1151,10 @@ async def initialize_server() -> MCPConfig:
     parser.add_argument(
         '--model', help=f'Model name to use with the LLM client. (default: {DEFAULT_LLM_MODEL})'
     )
+    parser.add_argument(
+        '--small-model',
+        help=f'Small model name to use with the LLM client. (default: {SMALL_LLM_MODEL})',
+    )
     parser.add_argument(
         '--temperature',
         type=float,