Fix: Remove models/ prefix causing LiteLLM provider recognition failure

- Adds validation to prevent models/ prefix in model names
- Adds tests for model name validation
- Ensures correct model name format for LiteLLM provider recognition

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

src/crewai/llm.py

@@ -142,6 +142,13 @@ class LLM:
         reasoning_effort: Optional[Literal["none", "low", "medium", "high"]] = None,
         **kwargs,
     ):
+        # Validate model name
+        if isinstance(model, str) and model.startswith('models/'):
+            raise ValueError(
+                'Model name should not start with "models/". '
+                'Use the provider prefix instead (e.g., "gemini/model-name").'
+            )
+        
         self.model = model
         self.timeout = timeout
         self.temperature = temperature

tests/llm_test.py

@@ -252,6 +252,17 @@ def test_validate_call_params_no_response_format():
     llm._validate_call_params()
 
 
+def test_model_name_validation():
+    """Test that model names with 'models/' prefix are rejected."""
+    with pytest.raises(ValueError, match="should not start with \"models/\""):
+        LLM(model="models/gemini/gemini-1.5-pro")
+
+    # Valid model names should work
+    LLM(model="gemini/gemini-1.5-pro")
+    LLM(model="anthropic/claude-3-opus-20240229-v1:0")
+    LLM(model="openai/gpt-4")
+
+
 @pytest.mark.vcr(filter_headers=["authorization"])
 def test_o3_mini_reasoning_effort_high():
     llm = LLM(