diff --git a/biochatter/llm_connect/llmlite.py b/biochatter/llm_connect/llmlite.py
index ef95fadc..1a844759 100644
--- a/biochatter/llm_connect/llmlite.py
+++ b/biochatter/llm_connect/llmlite.py
@@ -65,12 +65,12 @@ def __init__(
         self.ca_model_name = model_name
         self._update_token_usage = update_token_usage
 
-    def get_litellm_object(self, api_key: str, model: str) -> ChatLiteLLM:
+    def _get_litellm_object(self, api_key: str | None, model: str) -> ChatLiteLLM:
         """Get a LiteLLM object for the specified model and API key.
 
         Args:
         ----
-            api_key (str): The API key for the LLM provider.
+            api_key (str | None): The API key for the LLM provider.
             model (str): The name of the model to use.
 
         Returns:
@@ -159,11 +159,11 @@ def set_api_key(self, api_key: str, user: str | None = None) -> bool:
             if self.ca_model_name is None:
                 raise ValueError("Correction Model name is not set.")
 
-            self.chat = self.get_litellm_object(api_key, self.model_name)
+            self.chat = self._get_litellm_object(api_key, self.model_name)
             if self.chat is None:
                 raise TypeError("Failed to intialize primary agent chat object.")
 
-            self.ca_chat = self.get_litellm_object(api_key, self.ca_model_name)
+            self.ca_chat = self._get_litellm_object(api_key, self.ca_model_name)
             if self.ca_chat is None:
                 raise TypeError("Failed to intialize correcting agent chat object.")
 
diff --git a/test/test_llm_connect/test_litellm_conversation.py b/test/test_llm_connect/test_litellm_conversation.py
index 3d34880a..cda4d59f 100644
--- a/test/test_llm_connect/test_litellm_conversation.py
+++ b/test/test_llm_connect/test_litellm_conversation.py
@@ -26,7 +26,7 @@
 
 @patch("biochatter.llm_connect.llmlite.ChatLiteLLM")
 def test_get_litellm_object_valid(mock_chatlite):
-    """Test that get_litellm_object calls ChatLiteLLM with the correct parameters
+    """Test that _get_litellm_object calls ChatLiteLLM with the correct parameters
     for a supported model.
     """
     conv = LiteLLMConversation(model_name="gpt-3.5-turbo", prompts={})
@@ -35,7 +35,7 @@ def test_get_litellm_object_valid(mock_chatlite):
     dummy_api_key = "dummy_key"
     # Patch get_model_max_tokens to return a dummy token limit.
     conv.get_model_max_tokens = MagicMock(return_value=4096)
-    result = conv.get_litellm_object(dummy_api_key, "gpt-3.5-turbo")
+    result = conv._get_litellm_object(dummy_api_key, "gpt-3.5-turbo")
     mock_chatlite.assert_called_with(
         temperature=0, openai_api_key=dummy_api_key, max_token=4096, model_name="gpt-3.5-turbo"
     )
@@ -53,7 +53,7 @@ def test_get_litellm_object_unsupported_model(mock_chatlite, dummy_api_key="dumm
     mock_chatlite.return_value = dummy_instance
     uc.get_model_max_tokens = MagicMock(side_effect=ValueError("Unsupported model: unknown-model"))
 
-    result = uc.get_litellm_object(dummy_api_key, "unknown-model")
+    result = uc._get_litellm_object(dummy_api_key, "unknown-model")
 
     mock_chatlite.assert_called_with(
         temperature=0,
@@ -64,7 +64,33 @@ def test_get_litellm_object_unsupported_model(mock_chatlite, dummy_api_key="dumm
     assert result == dummy_instance
 
 
-@patch.object(LiteLLMConversation, "get_litellm_object")
+def test_get_litellm_object_none_api_key_raises():
+    """Test that _get_litellm_object rejects a None api_key at runtime.
+
+    Real callers can pass one (podcast.py sources it from os.getenv, which
+    returns None when the env var is unset), and the docstring documents
+    this as a ValueError, not a TypeError.
+    """
+    conv = LiteLLMConversation(model_name="gpt-3.5-turbo", prompts={})
+
+    with pytest.raises(ValueError, match="API key must not be None"):
+        conv._get_litellm_object(None, "gpt-3.5-turbo")
+
+
+def test_get_litellm_object_api_key_annotation_allows_none():
+    """Test that the api_key annotation allows None.
+
+    This must match the ValueError-on-None branch the method actually
+    implements (issue #306: the annotation previously declared a plain
+    `str`, which made that branch statically unreachable).
+    """
+    import typing
+
+    hints = typing.get_type_hints(LiteLLMConversation._get_litellm_object)
+    assert type(None) in typing.get_args(hints["api_key"])
+
+
+@patch.object(LiteLLMConversation, "_get_litellm_object")
 def test_set_api_key_success(mock_get_llm):
     """Test that set_api_key assigns chat and ca_chat correctly on success."""
     dummy_api_key = "dummy_key"
@@ -83,9 +109,9 @@ def test_set_api_key_success(mock_get_llm):
     assert uc.user == "test_user"
 
 
-@patch.object(LiteLLMConversation, "get_litellm_object")
+@patch.object(LiteLLMConversation, "_get_litellm_object")
 def test_set_api_key_failure(mock_get_llm, dummy_api_key="dummy_key"):
-    """Test that if get_litellm_object throws an exception, set_api_key returns False
+    """Test that if _get_litellm_object throws an exception, set_api_key returns False
     and does not initialize chat attributes.
     """
     mock_get_llm.side_effect = ValueError("Invalid API key")