Fix #2645: Remove 'models/' prefix from LLM model names

Co-Authored-By: Joe Moura <joao@crewai.com>

src/crewai/agent.py

@@ -196,9 +196,11 @@ class Agent(BaseAgent):
         else:
             # For any other type, attempt to extract relevant attributes
             llm_params = {
-                "model": getattr(self.llm, "model_name", None)
-                or getattr(self.llm, "deployment_name", None)
-                or str(self.llm),
+                "model": self._normalize_model_name(
+                    getattr(self.llm, "model_name", None)
+                    or getattr(self.llm, "deployment_name", None)
+                    or str(self.llm)
+                ),
                 "temperature": getattr(self.llm, "temperature", None),
                 "max_tokens": getattr(self.llm, "max_tokens", None),
                 "logprobs": getattr(self.llm, "logprobs", None),
@@ -534,5 +536,14 @@ class Agent(BaseAgent):
     def __tools_names(tools) -> str:
         return ", ".join([t.name for t in tools])
 
+    def _normalize_model_name(self, model_name):
+        """
+        Normalize the model name by removing any 'models/' prefix.
+        This fixes the issue with ChatGoogleGenerativeAI and potentially other LLM providers.
+        """
+        if model_name and isinstance(model_name, str) and model_name.startswith("models/"):
+            return model_name[7:]  # Remove "models/" prefix
+        return model_name
+
     def __repr__(self):
         return f"Agent(role={self.role}, goal={self.goal}, backstory={self.backstory})"

src/crewai/agents/crew_agent_executor.py

@@ -3,7 +3,6 @@ import re
 from dataclasses import dataclass
 from typing import Any, Dict, List, Union
 
-from litellm import AuthenticationError as LiteLLMAuthenticationError
 from crewai.agents.agent_builder.base_agent import BaseAgent
 from crewai.agents.agent_builder.base_agent_executor_mixin import CrewAgentExecutorMixin
 from crewai.agents.parser import (
@@ -198,19 +197,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
             return self._invoke_loop(formatted_answer)
 
         except Exception as e:
-            if isinstance(e, LiteLLMAuthenticationError):
-                self._logger.log(
-                    level="error",
-                    message="Authentication error with litellm occurred. Please check your API key and configuration.",
-                    color="red",
-                )
-                self._logger.log(
-                    level="error",
-                    message=f"Error details: {str(e)}",
-                    color="red",
-                )
-                raise e
-            elif LLMContextLengthExceededException(str(e))._is_context_limit_error(
+            if LLMContextLengthExceededException(str(e))._is_context_limit_error(
                 str(e)
             ):
                 self._handle_context_length()

tests/agent_test.py

@@ -1308,115 +1308,6 @@ def test_llm_call_with_error():
         llm.call(messages)
 
 
-@pytest.mark.vcr(filter_headers=["authorization"])
-def test_litellm_auth_error_handling():
-    """Test that LiteLLM authentication errors are handled correctly and not retried."""
-    from litellm import AuthenticationError as LiteLLMAuthenticationError
-
-    # Create an agent with a mocked LLM and max_retry_limit=0
-    agent = Agent(
-        role="test role",
-        goal="test goal",
-        backstory="test backstory",
-        llm=LLM(model="gpt-4"),
-        max_retry_limit=0,  # Disable retries for authentication errors
-        max_iter=1,  # Limit to one iteration to prevent multiple calls
-    )
-
-    # Create a task
-    task = Task(
-        description="Test task",
-        expected_output="Test output",
-        agent=agent,
-    )
-
-    # Mock the LLM call to raise LiteLLMAuthenticationError
-    with (
-        patch.object(LLM, "call") as mock_llm_call,
-        pytest.raises(LiteLLMAuthenticationError, match="Invalid API key"),
-    ):
-        mock_llm_call.side_effect = LiteLLMAuthenticationError(
-            message="Invalid API key",
-            llm_provider="openai",
-            model="gpt-4"
-        )
-        agent.execute_task(task)
-
-    # Verify the call was only made once (no retries)
-    mock_llm_call.assert_called_once()
-
-
-@pytest.mark.vcr(filter_headers=["authorization"])
-def test_crew_agent_executor_litellm_auth_error():
-    """Test that CrewAgentExecutor properly identifies and handles LiteLLM authentication errors."""
-    from litellm import AuthenticationError as LiteLLMAuthenticationError
-    from crewai.utilities import Logger
-    from crewai.agents.tools_handler import ToolsHandler
-
-    # Create an agent and executor with max_retry_limit=0
-    agent = Agent(
-        role="test role",
-        goal="test goal",
-        backstory="test backstory",
-        llm=LLM(model="gpt-4"),
-        max_retry_limit=0,  # Disable retries for authentication errors
-    )
-    task = Task(
-        description="Test task",
-        expected_output="Test output",
-        agent=agent,
-    )
-    
-    # Create executor with all required parameters
-    executor = CrewAgentExecutor(
-        agent=agent,
-        task=task,
-        llm=agent.llm,
-        crew=None,
-        prompt={
-            "system": "You are a test agent",
-            "user": "Execute the task: {input}"
-        },
-        max_iter=5,
-        tools=[],
-        tools_names="",
-        stop_words=[],
-        tools_description="",
-        tools_handler=ToolsHandler(),
-    )
-
-    # Mock the LLM call to raise LiteLLMAuthenticationError
-    with (
-        patch.object(LLM, "call") as mock_llm_call,
-        patch.object(Logger, "log") as mock_logger,
-        pytest.raises(LiteLLMAuthenticationError, match="Invalid API key"),
-    ):
-        mock_llm_call.side_effect = LiteLLMAuthenticationError(
-            message="Invalid API key",
-            llm_provider="openai",
-            model="gpt-4"
-        )
-        executor.invoke({
-            "input": "test input",
-            "tool_names": "",  # Required template variable
-            "tools": "",       # Required template variable
-        })
-
-    # Verify error handling
-    mock_logger.assert_any_call(
-        level="error",
-        message="Authentication error with litellm occurred. Please check your API key and configuration.",
-        color="red",
-    )
-    mock_logger.assert_any_call(
-        level="error",
-        message="Error details: litellm.AuthenticationError: Invalid API key",
-        color="red",
-    )
-    # Verify the call was only made once (no retries)
-    mock_llm_call.assert_called_once()
-
-
 @pytest.mark.vcr(filter_headers=["authorization"])
 def test_handle_context_length_exceeds_limit():
     agent = Agent(

tests/test_agent_model_name.py

@@ -0,0 +1,39 @@
+import pytest
+from unittest.mock import MagicMock, patch
+from crewai import Agent
+from crewai.llm import LLM
+
+
+def test_normalize_model_name_method():
+    """Test that the _normalize_model_name method correctly handles model names with 'models/' prefix"""
+    agent = Agent(
+        role="Test Agent",
+        goal="Test goal",
+        backstory="Test backstory",
+        llm="gpt-4"
+    )
+    
+    model_with_prefix = "models/gemini/gemini-1.5-flash"
+    normalized_name = agent._normalize_model_name(model_with_prefix)
+    assert normalized_name == "gemini/gemini-1.5-flash"
+    
+    regular_model = "gpt-4"
+    assert agent._normalize_model_name(regular_model) == "gpt-4"
+    
+    assert agent._normalize_model_name(None) is None
+    
+    assert agent._normalize_model_name(123) == 123
+
+
+def test_agent_with_regular_model_name():
+    """Test that the Agent class doesn't modify normal model names"""
+    with patch('crewai.agent.LLM') as mock_llm:
+        agent = Agent(
+            role="Test Agent",
+            goal="Test goal",
+            backstory="Test backstory",
+            llm="gpt-4"
+        )
+        
+        args, kwargs = mock_llm.call_args
+        assert kwargs["model"] == "gpt-4"