[LLM translation] Refactor Anthropic Configurations and Add Support for anthropic_beta Headers

docs/my-website/docs/providers/bedrock.md

@@ -584,6 +584,150 @@ Same as [Anthropic API response](../providers/anthropic#usage---thinking--reason
 Same as [Anthropic API response](../providers/anthropic#usage---thinking--reasoning_content).
 
 
+## Usage - Anthropic Beta Features
+
+LiteLLM supports Anthropic's beta features on AWS Bedrock through the `anthropic-beta` header. This enables access to experimental features like:
+
+- **1M Context Window** - Up to 1 million tokens of context (Claude Sonnet 4)
+- **Computer Use Tools** - AI that can interact with computer interfaces
+- **Token-Efficient Tools** - More efficient tool usage patterns  
+- **Extended Output** - Up to 128K output tokens
+- **Enhanced Thinking** - Advanced reasoning capabilities
+
+### Supported Beta Features
+
+| Beta Feature | Header Value | Compatible Models | Description |
+|--------------|-------------|------------------|-------------|
+| 1M Context Window | `context-1m-2025-08-07` | Claude Sonnet 4 | Enable 1 million token context window |
+| Computer Use (Latest) | `computer-use-2025-01-24` | Claude 3.7 Sonnet | Latest computer use tools |
+| Computer Use (Legacy) | `computer-use-2024-10-22` | Claude 3.5 Sonnet v2 | Computer use tools for Claude 3.5 |
+| Token-Efficient Tools | `token-efficient-tools-2025-02-19` | Claude 3.7 Sonnet | More efficient tool usage |
+| Interleaved Thinking | `interleaved-thinking-2025-05-14` | Claude 4 models | Enhanced thinking capabilities |
+| Extended Output | `output-128k-2025-02-19` | Claude 3.7 Sonnet | Up to 128K output tokens |
+| Developer Thinking | `dev-full-thinking-2025-05-14` | Claude 4 models | Raw thinking mode for developers |
+
+<Tabs>
+<TabItem value="sdk" label="SDK">
+
+**Single Beta Feature**
+
+```python
+from litellm import completion
+import os
+
+# set env
+os.environ["AWS_ACCESS_KEY_ID"] = ""
+os.environ["AWS_SECRET_ACCESS_KEY"] = ""
+os.environ["AWS_REGION_NAME"] = ""
+
+# Use 1M context window with Claude Sonnet 4
+response = completion(
+    model="bedrock/anthropic.claude-sonnet-4-20250115-v1:0",
+    messages=[{"role": "user", "content": "Hello! Testing 1M context window."}],
+    max_tokens=100,
+    extra_headers={
+        "anthropic-beta": "context-1m-2025-08-07"  # 👈 Enable 1M context
+    }
+)
+```
+
+**Multiple Beta Features**
+
+```python
+from litellm import completion
+
+# Combine multiple beta features (comma-separated)
+response = completion(
+    model="bedrock/converse/anthropic.claude-3-5-sonnet-20241022-v2:0",
+    messages=[{"role": "user", "content": "Testing multiple beta features"}],
+    max_tokens=100,
+    extra_headers={
+        "anthropic-beta": "computer-use-2024-10-22,context-1m-2025-08-07"
+    }
+)
+```
+
+**Computer Use Tools with Beta Features**
+
+```python
+from litellm import completion
+
+# Computer use tools automatically add computer-use-2024-10-22
+# You can add additional beta features
+response = completion(
+    model="bedrock/converse/anthropic.claude-3-5-sonnet-20241022-v2:0",
+    messages=[{"role": "user", "content": "Take a screenshot"}],
+    tools=[{
+        "type": "computer_20241022",
+        "name": "computer",
+        "display_width_px": 1920,
+        "display_height_px": 1080
+    }],
+    extra_headers={
+        "anthropic-beta": "context-1m-2025-08-07"  # Additional beta feature
+    }
+)
+```
+
+</TabItem>
+<TabItem value="proxy" label="PROXY">
+
+**Set on YAML Config**
+
+```yaml
+model_list:
+  - model_name: claude-sonnet-4-1m
+    litellm_params:
+      model: bedrock/anthropic.claude-sonnet-4-20250115-v1:0
+      extra_headers:
+        anthropic-beta: "context-1m-2025-08-07"  # 👈 Enable 1M context
+
+  - model_name: claude-computer-use
+    litellm_params:
+      model: bedrock/converse/anthropic.claude-3-5-sonnet-20241022-v2:0
+      extra_headers:
+        anthropic-beta: "computer-use-2024-10-22,context-1m-2025-08-07"
+
+general_settings:
+  forward_client_headers_to_llm_api: true  # 👈 Required for client-side header forwarding
+```
+
+**Set on Request**
+
+```python
+import openai
+
+client = openai.OpenAI(
+    api_key="anything",
+    base_url="http://0.0.0.0:4000"
+)
+
+response = client.chat.completions.create(
+    model="claude-sonnet-4-1m",
+    messages=[{
+        "role": "user", 
+        "content": "Testing 1M context window"
+    }],
+    extra_headers={
+        "anthropic-beta": "context-1m-2025-08-07"
+    }
+)
+```
+
+:::info
+**For client-side header forwarding**: When using the proxy and sending `anthropic-beta` headers from the client (like the OpenAI SDK), you need to enable `forward_client_headers_to_llm_api: true` in your proxy's `general_settings`. This tells the proxy to extract headers from HTTP requests and forward them to the underlying LLM provider.
+:::
+
+</TabItem>
+</Tabs>
+
+:::info
+
+Beta features may require special access or permissions in your AWS account. Some features are only available in specific AWS regions. Check the [AWS Bedrock documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-anthropic-claude-messages-request-response.html) for availability and access requirements.
+
+:::
+
+
 ## Usage - Structured Output / JSON mode 
 
 <Tabs>

litellm/__init__.py

@@ -1041,7 +1041,7 @@ def add_known_models():
     AnthropicMessagesConfig,
 )
 from .llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation import (
-    AmazonAnthropicClaude3MessagesConfig,
+    AmazonAnthropicClaudeMessagesConfig,
 )
 from .llms.together_ai.chat import TogetherAIConfig
 from .llms.together_ai.completion.transformation import TogetherAITextCompletionConfig
@@ -1101,7 +1101,7 @@ def add_known_models():
     AmazonAnthropicConfig,
 )
 from .llms.bedrock.chat.invoke_transformations.anthropic_claude3_transformation import (
-    AmazonAnthropicClaude3Config,
+    AmazonAnthropicClaudeConfig,
 )
 from .llms.bedrock.chat.invoke_transformations.amazon_cohere_transformation import (
     AmazonCohereConfig,

litellm/llms/bedrock/chat/converse_handler.py

@@ -119,6 +119,7 @@ async def async_streaming(
             messages=messages,
             optional_params=optional_params,
             litellm_params=litellm_params,
+            headers=headers,
         )
         data = json.dumps(request_data)
 
@@ -185,8 +186,10 @@ async def async_completion(
             messages=messages,
             optional_params=optional_params,
             litellm_params=litellm_params,
+            headers=headers,
         )
         data = json.dumps(request_data)
+
         prepped = self.get_request_headers(
             credentials=credentials,
             aws_region_name=litellm_params.get("aws_region_name") or "us-west-2",
@@ -390,8 +393,10 @@ def completion(  # noqa: PLR0915
             messages=messages,
             optional_params=optional_params,
             litellm_params=litellm_params,
+            headers=extra_headers,
         )
         data = json.dumps(_data)
+
         prepped = self.get_request_headers(
             credentials=credentials,
             aws_region_name=aws_region_name,

litellm/llms/bedrock/chat/converse_transformation.py

@@ -47,7 +47,7 @@
 )
 from litellm.utils import add_dummy_tool, has_tool_call_blocks, supports_reasoning
 
-from ..common_utils import BedrockError, BedrockModelInfo, get_bedrock_tool_name
+from ..common_utils import BedrockError, BedrockModelInfo, get_bedrock_tool_name, get_anthropic_beta_from_headers
 
 # Computer use tool prefixes supported by Bedrock
 BEDROCK_COMPUTER_USE_TOOLS = [
@@ -593,12 +593,14 @@ def _handle_top_k_value(self, model: str, inference_params: dict) -> dict:
 
         return {}
 
+
     def _transform_request_helper(
         self,
         model: str,
         system_content_blocks: List[SystemContentBlock],
         optional_params: dict,
         messages: Optional[List[AllMessageValues]] = None,
+        headers: Optional[dict] = None,
     ) -> CommonRequestObject:
         ## VALIDATE REQUEST
         """
@@ -651,6 +653,12 @@ def _transform_request_helper(
         # Initialize bedrock_tools
         bedrock_tools: List[ToolBlock] = []
 
+        # Collect anthropic_beta values from user headers
+        anthropic_beta_list = []
+        if headers:
+            user_betas = get_anthropic_beta_from_headers(headers)
+            anthropic_beta_list.extend(user_betas)
+
         # Only separate tools if computer use tools are actually present
         if original_tools and self.is_computer_use_tool_used(original_tools, model):
             # Separate computer use tools from regular function tools
@@ -663,14 +671,25 @@ def _transform_request_helper(
 
             # Add computer use tools and anthropic_beta if needed (only when computer use tools are present)
             if computer_use_tools:
-                additional_request_params["anthropic_beta"] = ["computer-use-2024-10-22"]
+                anthropic_beta_list.append("computer-use-2024-10-22")
                 # Transform computer use tools to proper Bedrock format
                 transformed_computer_tools = self._transform_computer_use_tools(computer_use_tools)
                 additional_request_params["tools"] = transformed_computer_tools
         else:
             # No computer use tools, process all tools as regular tools
             bedrock_tools = _bedrock_tools_pt(original_tools)
 
+        # Set anthropic_beta in additional_request_params if we have any beta features
+        if anthropic_beta_list:
+            # Remove duplicates while preserving order
+            unique_betas = []
+            seen = set()
+            for beta in anthropic_beta_list:
+                if beta not in seen:
+                    unique_betas.append(beta)
+                    seen.add(beta)
+            additional_request_params["anthropic_beta"] = unique_betas
+
         bedrock_tool_config: Optional[ToolConfigBlock] = None
         if len(bedrock_tools) > 0:
             tool_choice_values: ToolChoiceValuesBlock = inference_params.pop(
@@ -708,6 +727,7 @@ async def _async_transform_request(
         messages: List[AllMessageValues],
         optional_params: dict,
         litellm_params: dict,
+        headers: Optional[dict] = None,
     ) -> RequestObject:
         messages, system_content_blocks = self._transform_system_message(messages)
         ## TRANSFORMATION ##
@@ -717,6 +737,7 @@ async def _async_transform_request(
             system_content_blocks=system_content_blocks,
             optional_params=optional_params,
             messages=messages,
+            headers=headers,
         )
 
         bedrock_messages = (
@@ -747,6 +768,7 @@ def transform_request(
                 messages=messages,
                 optional_params=optional_params,
                 litellm_params=litellm_params,
+                headers=headers,
             ),
         )
 
@@ -756,6 +778,7 @@ def _transform_request(
         messages: List[AllMessageValues],
         optional_params: dict,
         litellm_params: dict,
+        headers: Optional[dict] = None,
     ) -> RequestObject:
         messages, system_content_blocks = self._transform_system_message(messages)
 
@@ -764,6 +787,7 @@ def _transform_request(
             system_content_blocks=system_content_blocks,
             optional_params=optional_params,
             messages=messages,
+            headers=headers,
         )
 
         ## TRANSFORMATION ##

litellm/llms/bedrock/chat/invoke_handler.py

@@ -831,7 +831,7 @@ def completion(  # noqa: PLR0915
                     model=model, messages=messages, custom_llm_provider="anthropic_xml"
                 )  # type: ignore
                 ## LOAD CONFIG
-                config = litellm.AmazonAnthropicClaude3Config.get_config()
+                config = litellm.AmazonAnthropicClaudeConfig.get_config()
                 for k, v in config.items():
                     if (
                         k not in inference_params

litellm/llms/bedrock/chat/invoke_transformations/anthropic_claude3_transformation.py

@@ -6,6 +6,7 @@
 from litellm.llms.bedrock.chat.invoke_transformations.base_invoke_transformation import (
     AmazonInvokeConfig,
 )
+from litellm.llms.bedrock.common_utils import get_anthropic_beta_from_headers
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.utils import ModelResponse
 
@@ -17,13 +18,22 @@
     LiteLLMLoggingObj = Any
 
 
-class AmazonAnthropicClaude3Config(AmazonInvokeConfig, AnthropicConfig):
+class AmazonAnthropicClaudeConfig(AmazonInvokeConfig, AnthropicConfig):
     """
     Reference:
         https://us-west-2.console.aws.amazon.com/bedrock/home?region=us-west-2#/providers?model=claude
         https://docs.anthropic.com/claude/docs/models-overview#model-comparison
+        https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-anthropic-claude-messages-request-response.html
 
-    Supported Params for the Amazon / Anthropic Claude 3 models:
+    Supported Params for the Amazon / Anthropic Claude models (Claude 3, Claude 4, etc.):
+    Supports anthropic_beta parameter for beta features like:
+    - computer-use-2025-01-24 (Claude 3.7 Sonnet)
+    - computer-use-2024-10-22 (Claude 3.5 Sonnet v2)
+    - token-efficient-tools-2025-02-19 (Claude 3.7 Sonnet)
+    - interleaved-thinking-2025-05-14 (Claude 4 models)
+    - output-128k-2025-02-19 (Claude 3.7 Sonnet)
+    - dev-full-thinking-2025-05-14 (Claude 4 models)
+    - context-1m-2025-08-07 (Claude Sonnet 4)
     """
 
     anthropic_version: str = "bedrock-2023-05-31"
@@ -50,6 +60,7 @@ def map_openai_params(
             drop_params,
         )
 
+
     def transform_request(
         self,
         model: str,
@@ -72,6 +83,11 @@ def transform_request(
         if "anthropic_version" not in _anthropic_request:
             _anthropic_request["anthropic_version"] = self.anthropic_version
 
+        # Handle anthropic_beta from user headers
+        anthropic_beta_list = get_anthropic_beta_from_headers(headers)
+        if anthropic_beta_list:
+            _anthropic_request["anthropic_beta"] = anthropic_beta_list
+
         return _anthropic_request
 
     def transform_response(

litellm/llms/bedrock/chat/invoke_transformations/base_invoke_transformation.py

@@ -190,13 +190,15 @@ def transform_request(
                     ] = True  # cohere requires stream = True in inference params
                 request_data = {"prompt": prompt, **inference_params}
         elif provider == "anthropic":
-            return litellm.AmazonAnthropicClaude3Config().transform_request(
+            transformed_request = litellm.AmazonAnthropicClaudeConfig().transform_request(
                 model=model,
                 messages=messages,
                 optional_params=optional_params,
                 litellm_params=litellm_params,
                 headers=headers,
             )
+
+            return transformed_request
         elif provider == "nova":
             return litellm.AmazonInvokeNovaConfig().transform_request(
                 model=model,
@@ -293,7 +295,7 @@ def transform_response(  # noqa: PLR0915
                         completion_response["generations"][0]["finish_reason"]
                     )
             elif provider == "anthropic":
-                return litellm.AmazonAnthropicClaude3Config().transform_response(
+                return litellm.AmazonAnthropicClaudeConfig().transform_response(
                     model=model,
                     raw_response=raw_response,
                     model_response=model_response,