(Bug)Migrate Anthropic provider to azure ai

docs/my-website/blog/anthropic_opus_4_5_and_advanced_features/index.md

@@ -327,6 +327,81 @@ curl --location 'http://0.0.0.0:4000/v1/messages' \
 </TabItem>
 </Tabs>
 
+## Usage - Azure Anthropic (Azure Foundry Claude)
+
+LiteLLM funnels Azure Claude deployments through the `azure_ai/` provider so Claude Opus models on Azure Foundry keep working with Tool Search, Effort, streaming, and the rest of the advanced feature set. Point `AZURE_AI_API_BASE` to `https://<resource>.services.ai.azure.com/anthropic` (LiteLLM appends `/v1/messages` automatically) and authenticate with `AZURE_AI_API_KEY` or an Azure AD token.
+
+<Tabs>
+<TabItem value="sdk" label="LiteLLM Python SDK">
+
+```python
+import os
+from litellm import completion
+
+# Configure Azure credentials
+os.environ["AZURE_AI_API_KEY"] = "your-azure-ai-api-key"
+os.environ["AZURE_AI_API_BASE"] = "https://my-resource.services.ai.azure.com/anthropic"
+
+response = completion(
+    model="azure_ai/claude-opus-4-1",
+    messages=[{"role": "user", "content": "Explain how Azure Anthropic hosts Claude Opus differently from the public Anthropic API."}],
+    max_tokens=1200,
+    temperature=0.7,
+    stream=True,
+)
+
+for chunk in response:
+    if chunk.choices[0].delta.content:
+        print(chunk.choices[0].delta.content, end="", flush=True)
+```
+
+</TabItem>
+<TabItem value="proxy" label="LiteLLM Proxy">
+
+**1. Set environment variables**
+
+```bash
+export AZURE_AI_API_KEY="your-azure-ai-api-key"
+export AZURE_AI_API_BASE="https://my-resource.services.ai.azure.com/anthropic"
+```
+
+**2. Configure the proxy**
+
+```yaml
+model_list:
+  - model_name: claude-4-azure
+    litellm_params:
+      model: azure_ai/claude-opus-4-1
+      api_key: os.environ/AZURE_AI_API_KEY
+      api_base: os.environ/AZURE_AI_API_BASE
+```
+
+**3. Start LiteLLM**
+
+```bash
+litellm --config /path/to/config.yaml
+```
+
+**4. Test the Azure Claude route**
+
+```bash
+curl --location 'http://0.0.0.0:4000/chat/completions' \
+  --header 'Content-Type: application/json' \
+  --header 'Authorization: Bearer $LITELLM_KEY' \
+  --data '{
+    "model": "claude-4-azure",
+    "messages": [
+      {
+        "role": "user",
+        "content": "How do I use Claude Opus 4 via Azure Anthropic in LiteLLM?"
+      }
+    ],
+    "max_tokens": 1024
+  }'
+```
+
+</TabItem>
+</Tabs>
 
 
 ## Tool Search {#tool-search}

docs/my-website/docs/providers/azure/azure_anthropic.md

@@ -16,7 +16,7 @@ Azure Foundry supports the following Claude models:
 | Property | Details |
 |-------|-------|
 | Description | Claude models deployed via Microsoft Azure Foundry. Uses the same API as Anthropic's Messages API but with Azure authentication. |
-| Provider Route on LiteLLM | `azure/` (add this prefix to Claude model names - e.g. `azure/claude-sonnet-4-5`) |
+| Provider Route on LiteLLM | `azure_ai/` (add this prefix to Claude model names - e.g. `azure_ai/claude-sonnet-4-5`) |
 | Provider Doc | [Azure Foundry Claude Models ↗](https://learn.microsoft.com/en-us/azure/ai-services/foundry-models/claude) |
 | API Endpoint | `https://<resource-name>.services.ai.azure.com/anthropic/v1/messages` |
 | Supported Endpoints | `/chat/completions`, `/anthropic/v1/messages`|
@@ -68,7 +68,7 @@ os.environ["AZURE_API_BASE"] = "https://<resource-name>.services.ai.azure.com/an
 
 # Make a completion request
 response = completion(
-    model="azure/claude-sonnet-4-5",
+    model="azure_ai/claude-sonnet-4-5",
     messages=[
         {"role": "user", "content": "What are 3 things to visit in Seattle?"}
     ],
@@ -85,7 +85,7 @@ print(response)
 import litellm
 
 response = litellm.completion(
-    model="azure/claude-sonnet-4-5",
+    model="azure_ai/claude-sonnet-4-5",
     api_base="https://<resource-name>.services.ai.azure.com/anthropic",
     api_key="your-azure-api-key",
     messages=[
@@ -101,7 +101,7 @@ response = litellm.completion(
 import litellm
 
 response = litellm.completion(
-    model="azure/claude-sonnet-4-5",
+    model="azure_ai/claude-sonnet-4-5",
     api_base="https://<resource-name>.services.ai.azure.com/anthropic",
     azure_ad_token="your-azure-ad-token",
     messages=[
@@ -117,7 +117,7 @@ response = litellm.completion(
 from litellm import completion
 
 response = completion(
-    model="azure/claude-sonnet-4-5",
+    model="azure_ai/claude-sonnet-4-5",
     messages=[
         {"role": "user", "content": "Write a short story"}
     ],
@@ -136,7 +136,7 @@ for chunk in response:
 from litellm import completion
 
 response = completion(
-    model="azure/claude-sonnet-4-5",
+    model="azure_ai/claude-sonnet-4-5",
     messages=[
         {"role": "user", "content": "What's the weather in Seattle?"}
     ],
@@ -181,7 +181,7 @@ export AZURE_API_BASE="https://<resource-name>.services.ai.azure.com/anthropic"
 model_list:
   - model_name: claude-sonnet-4-5
     litellm_params:
-      model: azure/claude-sonnet-4-5
+      model: azure_ai/claude-sonnet-4-5
       api_base: https://<resource-name>.services.ai.azure.com/anthropic
       api_key: os.environ/AZURE_API_KEY
 ```
@@ -331,7 +331,7 @@ os.environ["AZURE_API_BASE"] = "https://my-resource.services.ai.azure.com/anthro
 
 # Make a request
 response = completion(
-    model="azure/claude-sonnet-4-5",
+    model="azure_ai/claude-sonnet-4-5",
     messages=[
         {"role": "system", "content": "You are a helpful assistant."},
         {"role": "user", "content": "Explain quantum computing in simple terms."}
@@ -358,7 +358,7 @@ Or pass it directly:
 
 ```python
 response = completion(
-    model="azure/claude-sonnet-4-5",
+    model="azure_ai/claude-sonnet-4-5",
     api_base="https://<resource-name>.services.ai.azure.com/anthropic",
     # ...
 )

litellm/__init__.py

@@ -1120,7 +1120,7 @@ def add_known_models():
 from .llms.datarobot.chat.transformation import DataRobotConfig
 from .llms.anthropic.chat.transformation import AnthropicConfig
 from .llms.anthropic.common_utils import AnthropicModelInfo
-from .llms.azure.anthropic.transformation import AzureAnthropicConfig
+from .llms.azure_ai.anthropic.transformation import AzureAnthropicConfig
 from .llms.groq.stt.transformation import GroqSTTConfig
 from .llms.anthropic.completion.transformation import AnthropicTextConfig
 from .llms.triton.completion.transformation import TritonConfig

litellm/litellm_core_utils/get_llm_provider_logic.py

@@ -22,17 +22,16 @@ def _is_non_openai_azure_model(model: str) -> bool:
     return False
 
 
-def _is_azure_anthropic_model(model: str) -> Optional[str]:
+def _is_azure_claude_model(model: str) -> bool:
+    """
+    Check if a model name contains 'claude' (case-insensitive).
+    Used to detect Claude models that need Anthropic-specific handling.
+    """
     try:
-        model_parts = model.split("/", 1)
-        if len(model_parts) > 1:
-            model_name = model_parts[1].lower()
-            # Check if model name contains claude
-            if "claude" in model_name or model_name.startswith("claude"):
-                return model_parts[1]  # Return model name without "azure/" prefix
+        model_lower = model.lower()
+        return "claude" in model_lower or model_lower.startswith("claude")
     except Exception:
-        pass
-    return None
+        return False
 
 
 def handle_cohere_chat_model_custom_llm_provider(
@@ -136,11 +135,6 @@ def get_llm_provider(  # noqa: PLR0915
         # AZURE AI-Studio Logic - Azure AI Studio supports AZURE/Cohere
         # If User passes azure/command-r-plus -> we should send it to cohere_chat/command-r-plus
         if model.split("/", 1)[0] == "azure":
-            # Check if it's an Azure Anthropic model (claude models)
-            azure_anthropic_model = _is_azure_anthropic_model(model)
-            if azure_anthropic_model:
-                custom_llm_provider = "azure_anthropic"
-                return azure_anthropic_model, custom_llm_provider, dynamic_api_key, api_base
             if _is_non_openai_azure_model(model):
                 custom_llm_provider = "openai"
                 return model, custom_llm_provider, dynamic_api_key, api_base

litellm/llms/azure/anthropic/handler.py → litellm/llms/azure_ai/anthropic/handler.py

@@ -7,7 +7,6 @@
 
 import httpx
 
-import litellm
 from litellm.llms.anthropic.chat.handler import AnthropicChatCompletion
 from litellm.llms.custom_httpx.http_handler import (
     AsyncHTTPHandler,
@@ -55,7 +54,6 @@ def completion(
         Completion method that uses Azure authentication instead of Anthropic's x-api-key.
         All other logic is the same as AnthropicChatCompletion.
         """
-        from litellm.utils import ProviderConfigManager
 
         optional_params = copy.deepcopy(optional_params)
         stream = optional_params.pop("stream", None)
@@ -64,8 +62,10 @@ def completion(
         _is_function_call = False
         messages = copy.deepcopy(messages)
 
-        # Use AzureAnthropicConfig instead of AnthropicConfig
-        headers = AzureAnthropicConfig().validate_environment(
+        # Use AzureAnthropicConfig for both azure_anthropic and azure_ai Claude models
+        config = AzureAnthropicConfig()
+
+        headers = config.validate_environment(
             api_key=api_key,
             headers=headers,
             model=model,
@@ -74,15 +74,6 @@ def completion(
             litellm_params=litellm_params,
         )
 
-        config = ProviderConfigManager.get_provider_chat_config(
-            model=model,
-            provider=litellm.types.utils.LlmProviders(custom_llm_provider),
-        )
-        if config is None:
-            raise ValueError(
-                f"Provider config not found for model: {model} and provider: {custom_llm_provider}"
-            )
-
         data = config.transform_request(
             model=model,
             messages=messages,
@@ -183,7 +174,7 @@ def completion(
                 return CustomStreamWrapper(
                     completion_stream=completion_stream,
                     model=model,
-                    custom_llm_provider="azure_anthropic",
+                    custom_llm_provider="azure_ai",
                     logging_obj=logging_obj,
                     _response_headers=process_anthropic_headers(response_headers),
                 )

litellm/llms/azure/anthropic/transformation.py → litellm/llms/azure_ai/anthropic/transformation.py

@@ -21,7 +21,7 @@ class AzureAnthropicConfig(AnthropicConfig):
 
     @property
     def custom_llm_provider(self) -> Optional[str]:
-        return "azure_anthropic"
+        return "azure_ai"
 
     def validate_environment(
         self,
@@ -94,3 +94,29 @@ def validate_environment(
 
         return headers
 
+    def transform_request(
+        self,
+        model: str,
+        messages: List[AllMessageValues],
+        optional_params: dict,
+        litellm_params: dict,
+        headers: dict,
+    ) -> dict:
+        """
+        Transform request using parent AnthropicConfig, then remove extra_body if present.
+        Azure Anthropic doesn't support extra_body parameter.
+        """
+        # Call parent transform_request
+        data = super().transform_request(
+            model=model,
+            messages=messages,
+            optional_params=optional_params,
+            litellm_params=litellm_params,
+            headers=headers,
+        )
+
+        # Remove extra_body if present (Azure Anthropic doesn't support it)
+        data.pop("extra_body", None)
+
+        return data
+