Fix Azure OpenAI authentication for models without azure/ prefix (fixes #2358)

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

src/crewai/llm.py

@@ -829,10 +829,18 @@ class LLM:
         Derives the custom_llm_provider from the model string.
         - For example, if the model is "openrouter/deepseek/deepseek-chat", returns "openrouter".
         - If the model is "gemini/gemini-1.5-pro", returns "gemini".
-        - If there is no '/', defaults to "openai".
+        - If the model is "azure/gpt-4", returns "azure".
+        - If Azure-specific parameters are provided (api_key, api_base, api_version), returns "azure".
+        - If there is no '/' and no Azure parameters, returns None.
         """
+        # Check if model explicitly has a provider prefix
         if "/" in self.model:
             return self.model.split("/")[0]
+        
+        # Check if all Azure parameters are present
+        if self.api_key and self.api_base and self.api_version:
+            return "azure"
+            
         return None
 
     def _validate_call_params(self) -> None:
@@ -842,7 +850,9 @@ class LLM:
         The custom_llm_provider is dynamically determined from the model:
           - E.g., "openrouter/deepseek/deepseek-chat" yields "openrouter"
           - "gemini/gemini-1.5-pro" yields "gemini"
-          - If no slash is present, "openai" is assumed.
+          - "azure/gpt-4" yields "azure"
+          - If Azure parameters (api_key, api_base, api_version) are present, "azure" is used
+          - If no slash is present and no Azure parameters, None is returned
         """
         provider = self._get_custom_llm_provider()
         if self.response_format is not None and not supports_response_schema(

test_azure_integration.py

@@ -0,0 +1,37 @@
+# test_azure_integration.py
+from src.crewai.llm import LLM
+
+# Test with Azure parameters but without azure/ prefix
+llm = LLM(
+    api_key='test_key',
+    api_base='test_base',
+    model='gpt-4o-mini-2024-07-18',
+    api_version='test_version'
+)
+
+# Print the detected provider
+provider = llm._get_custom_llm_provider()
+print(f"Detected provider: {provider}")
+print(f"Is Azure detected correctly: {provider == 'azure'}")
+
+# Prepare parameters that would be passed to LiteLLM
+params = llm._prepare_completion_params(messages=[{"role": "user", "content": "test"}])
+print(f"Parameters passed to LiteLLM: {params}")
+
+# Test with Azure parameters and with azure/ prefix for comparison
+llm_with_prefix = LLM(
+    api_key='test_key',
+    api_base='test_base',
+    model='azure/gpt-4o-mini-2024-07-18',
+    api_version='test_version'
+)
+
+# Print the detected provider
+provider_with_prefix = llm_with_prefix._get_custom_llm_provider()
+print(f"\nWith azure/ prefix:")
+print(f"Detected provider: {provider_with_prefix}")
+print(f"Is Azure detected correctly: {provider_with_prefix == 'azure'}")
+
+# Prepare parameters that would be passed to LiteLLM
+params_with_prefix = llm_with_prefix._prepare_completion_params(messages=[{"role": "user", "content": "test"}])
+print(f"Parameters passed to LiteLLM: {params_with_prefix}")

tests/llm_test.py

@@ -221,6 +221,20 @@ def test_get_custom_llm_provider_openai():
     llm = LLM(model="gpt-4")
     assert llm._get_custom_llm_provider() == None
 
+def test_get_custom_llm_provider_azure_with_prefix():
+    llm = LLM(model="azure/gpt-4")
+    assert llm._get_custom_llm_provider() == "azure"
+
+
+def test_get_custom_llm_provider_azure_without_prefix():
+    llm = LLM(
+        model="gpt-4",
+        api_key="test_key",
+        api_base="test_base",
+        api_version="test_version"
+    )
+    assert llm._get_custom_llm_provider() == "azure"
+
 
 def test_validate_call_params_supported():
     class DummyResponse(BaseModel):