feat(vertex_ai): Add exclude_anthropic_beta_values config option

litellm/interactions/http_handler.py

@@ -4,7 +4,6 @@
 This module handles the HTTP communication for the Google Interactions API.
 """
 
-import json
 from typing import (
     Any,
     AsyncIterator,
@@ -18,7 +17,6 @@
 import httpx
 
 import litellm
-from litellm._logging import verbose_logger
 from litellm.constants import request_timeout
 from litellm.interactions.streaming_iterator import (
     InteractionsAPIStreamingIterator,

litellm/interactions/streaming_iterator.py

@@ -8,11 +8,10 @@
 import asyncio
 import json
 from datetime import datetime
-from typing import Any, Dict, Iterator, Optional
+from typing import Any, Dict, Optional
 
 import httpx
 
-import litellm
 from litellm._logging import verbose_logger
 from litellm.constants import STREAM_SSE_DONE_STRING
 from litellm.litellm_core_utils.asyncify import run_async_function
@@ -22,7 +21,6 @@
 from litellm.litellm_core_utils.thread_pool_executor import executor
 from litellm.llms.base_llm.interactions.transformation import BaseInteractionsAPIConfig
 from litellm.types.interactions import (
-    InteractionsAPIResponse,
     InteractionsAPIStreamingResponse,
 )
 from litellm.utils import CustomStreamWrapper

litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/experimental_pass_through/transformation.py

@@ -50,15 +50,30 @@ def validate_anthropic_messages_environment(
             )
 
         headers["content-type"] = "application/json"
-
+
+        # Get exclusion list from litellm_params
+        exclude_anthropic_beta_values = litellm_params.get("exclude_anthropic_beta_values") or []
+
+        # Filter existing anthropic-beta header if present and exclusions are configured
+        if "anthropic-beta" in headers and exclude_anthropic_beta_values:
+            existing_betas = [b.strip() for b in headers["anthropic-beta"].split(",")]
+            filtered_betas = [b for b in existing_betas if b not in exclude_anthropic_beta_values]
+            if filtered_betas:
+                headers["anthropic-beta"] = ",".join(filtered_betas)
+            else:
+                del headers["anthropic-beta"]
+
         # Add web search beta header for Vertex AI only if not already set
         if "anthropic-beta" not in headers:
             tools = optional_params.get("tools", [])
             for tool in tools:
                 if isinstance(tool, dict) and tool.get("type", "").startswith(ANTHROPIC_HOSTED_TOOLS.WEB_SEARCH.value):
-                    headers["anthropic-beta"] = ANTHROPIC_BETA_HEADER_VALUES.WEB_SEARCH_2025_03_05.value
+                    web_search_beta = ANTHROPIC_BETA_HEADER_VALUES.WEB_SEARCH_2025_03_05.value
+                    # Only add if not in exclusion list
+                    if web_search_beta not in exclude_anthropic_beta_values:
+                        headers["anthropic-beta"] = web_search_beta
                     break
-        
+
         return headers, api_base
 
     def get_complete_url(

litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/transformation.py

@@ -84,9 +84,14 @@ def transform_request(
         if tool_search_used:
             beta_set.add("tool-search-tool-2025-10-19")  # Vertex requires this header for tool search
 
+        # Filter out excluded anthropic beta values if configured
+        exclude_anthropic_beta_values = litellm_params.get("exclude_anthropic_beta_values") or []
+        if exclude_anthropic_beta_values:
+            beta_set = {b for b in beta_set if b not in exclude_anthropic_beta_values}
+
         if beta_set:
             data["anthropic_beta"] = list(beta_set)
-        
+
         return data
 
     def transform_response(

litellm/proxy/google_endpoints/endpoints.py

@@ -1,6 +1,4 @@
-from typing import Optional
-
-from fastapi import APIRouter, Depends, HTTPException, Query, Request, Response
+from fastapi import APIRouter, Depends, HTTPException, Request, Response
 from fastapi.responses import ORJSONResponse, StreamingResponse
 
 from litellm.proxy._types import *

litellm/types/router.py

@@ -213,6 +213,9 @@ class GenericLiteLLMParams(CredentialLiteLLMParams, CustomPricingLiteLLMParams):
     vector_store_id: Optional[str] = None
     milvus_text_field: Optional[str] = None
 
+    # Anthropic Beta Header Filtering
+    exclude_anthropic_beta_values: Optional[List[str]] = None
+
     def __init__(
         self,
         custom_llm_provider: Optional[str] = None,
@@ -266,6 +269,8 @@ def __init__(
         s3_bucket_name: Optional[str] = None,
         s3_encryption_key_id: Optional[str] = None,
         gcs_bucket_name: Optional[str] = None,
+        # Anthropic Beta Header Filtering
+        exclude_anthropic_beta_values: Optional[List[str]] = None,
         **params,
     ):
         args = locals()

tests/test_litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_messages_config.py

@@ -98,3 +98,132 @@ def test_web_search_header_not_added_without_tool():
         # Assert that the anthropic-beta header is NOT present when no web search tool
         assert "anthropic-beta" not in updated_headers, \
             "anthropic-beta header should not be present without web search tool"
+
+
+class TestExcludeAnthropicBetaValuesPassThrough:
+    """Tests for exclude_anthropic_beta_values in pass-through handler"""
+
+    def test_exclude_filters_existing_header_values(self):
+        """Test that exclude_anthropic_beta_values filters values from existing anthropic-beta header"""
+        config = VertexAIPartnerModelsAnthropicMessagesConfig()
+        headers = {"anthropic-beta": "prompt-caching-2024-07-31,context-1m-2025-08-07"}
+        litellm_params = {
+            "vertex_ai_project": "test-project",
+            "vertex_ai_location": "us-central1",
+            "vertex_credentials": "{}",
+            "exclude_anthropic_beta_values": ["context-1m-2025-08-07"],
+        }
+        optional_params = {}
+
+        with patch.object(
+            config, "_ensure_access_token", return_value=("token", "test-project")
+        ), patch.object(
+            config, "get_complete_vertex_url", return_value="https://mock-url"
+        ):
+            updated_headers, api_base = config.validate_anthropic_messages_environment(
+                headers=headers,
+                model="claude-sonnet-4",
+                messages=[],
+                optional_params=optional_params,
+                litellm_params=litellm_params,
+                api_base=None,
+            )
+
+            # context-1m-2025-08-07 should be filtered out
+            assert "anthropic-beta" in updated_headers
+            assert "context-1m-2025-08-07" not in updated_headers["anthropic-beta"]
+            assert "prompt-caching-2024-07-31" in updated_headers["anthropic-beta"]
+
+    def test_exclude_removes_header_when_all_values_filtered(self):
+        """Test that anthropic-beta header is removed when all values are excluded"""
+        config = VertexAIPartnerModelsAnthropicMessagesConfig()
+        headers = {"anthropic-beta": "context-1m-2025-08-07"}
+        litellm_params = {
+            "vertex_ai_project": "test-project",
+            "vertex_ai_location": "us-central1",
+            "vertex_credentials": "{}",
+            "exclude_anthropic_beta_values": ["context-1m-2025-08-07"],
+        }
+        optional_params = {}
+
+        with patch.object(
+            config, "_ensure_access_token", return_value=("token", "test-project")
+        ), patch.object(
+            config, "get_complete_vertex_url", return_value="https://mock-url"
+        ):
+            updated_headers, api_base = config.validate_anthropic_messages_environment(
+                headers=headers,
+                model="claude-sonnet-4",
+                messages=[],
+                optional_params=optional_params,
+                litellm_params=litellm_params,
+                api_base=None,
+            )
+
+            # anthropic-beta header should be removed entirely
+            assert "anthropic-beta" not in updated_headers
+
+    def test_exclude_prevents_auto_added_web_search_header(self):
+        """Test that exclude_anthropic_beta_values can prevent auto-added web search header"""
+        config = VertexAIPartnerModelsAnthropicMessagesConfig()
+        headers = {}
+        litellm_params = {
+            "vertex_ai_project": "test-project",
+            "vertex_ai_location": "us-central1",
+            "vertex_credentials": "{}",
+            "exclude_anthropic_beta_values": ["web-search-2025-03-05"],
+        }
+        # Include web search tool that would normally add the header
+        optional_params = {
+            "tools": [
+                {"type": "web_search_20250305", "name": "web_search", "max_uses": 5}
+            ]
+        }
+
+        with patch.object(
+            config, "_ensure_access_token", return_value=("token", "test-project")
+        ), patch.object(
+            config, "get_complete_vertex_url", return_value="https://mock-url"
+        ):
+            updated_headers, api_base = config.validate_anthropic_messages_environment(
+                headers=headers,
+                model="claude-sonnet-4",
+                messages=[],
+                optional_params=optional_params,
+                litellm_params=litellm_params,
+                api_base=None,
+            )
+
+            # web-search header should NOT be added because it's in the exclude list
+            assert "anthropic-beta" not in updated_headers
+
+    def test_empty_exclude_list_preserves_headers(self):
+        """Test that empty exclude list doesn't filter any headers"""
+        config = VertexAIPartnerModelsAnthropicMessagesConfig()
+        headers = {"anthropic-beta": "prompt-caching-2024-07-31,context-1m-2025-08-07"}
+        litellm_params = {
+            "vertex_ai_project": "test-project",
+            "vertex_ai_location": "us-central1",
+            "vertex_credentials": "{}",
+            "exclude_anthropic_beta_values": [],
+        }
+        optional_params = {}
+
+        with patch.object(
+            config, "_ensure_access_token", return_value=("token", "test-project")
+        ), patch.object(
+            config, "get_complete_vertex_url", return_value="https://mock-url"
+        ):
+            updated_headers, api_base = config.validate_anthropic_messages_environment(
+                headers=headers,
+                model="claude-sonnet-4",
+                messages=[],
+                optional_params=optional_params,
+                litellm_params=litellm_params,
+                api_base=None,
+            )
+
+            # All values should be preserved
+            assert "anthropic-beta" in updated_headers
+            assert "prompt-caching-2024-07-31" in updated_headers["anthropic-beta"]
+            assert "context-1m-2025-08-07" in updated_headers["anthropic-beta"]