diff --git a/tests/test_empty_messages.py b/tests/test_empty_messages.py
index efffa294a..564fb65a1 100644
--- a/tests/test_empty_messages.py
+++ b/tests/test_empty_messages.py
@@ -1,11 +1,12 @@
-from unittest.mock import patch
+from unittest.mock import patch, MagicMock
 
 import pytest
 
 from crewai.llm import LLM
 
 
-def test_empty_messages_validation():
+@patch('crewai.llm.LLM._prepare_completion_params')
+def test_empty_messages_validation(mock_prepare):
     """
     Test that LLM.call() raises a ValueError when an empty messages list is passed.
     This prevents the IndexError in LiteLLM's ollama_pt() function.
@@ -17,6 +18,33 @@ def test_empty_messages_validation():
     
     with pytest.raises(ValueError, match="Messages list cannot be empty"):
         llm.call(messages=None)
+    
+    mock_prepare.assert_not_called()
+
+
+@patch('crewai.llm.LLM._prepare_completion_params')
+def test_empty_string_message(mock_prepare):
+    """
+    Test that LLM.call() raises a ValueError when an empty string message is passed.
+    """
+    llm = LLM(model="gpt-3.5-turbo")
+    
+    with pytest.raises(ValueError, match="Messages list cannot be empty"):
+        llm.call(messages="")
+    
+    mock_prepare.assert_not_called()
+
+
+@patch('crewai.llm.LLM._prepare_completion_params')
+def test_invalid_message_format(mock_prepare):
+    """
+    Test that LLM.call() raises a TypeError when a message with invalid format is passed.
+    """
+    mock_prepare.side_effect = TypeError("Invalid message format")
+    llm = LLM(model="gpt-3.5-turbo")
+    
+    with pytest.raises(TypeError, match="Invalid message format"):
+        llm.call(messages=[{}])
 
 
 @pytest.mark.vcr(filter_headers=["authorization"])