Fix: remove unsupported prompt-caching-scope-2026-01-05 header for vertex ai

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

@@ -7,6 +7,7 @@
 from litellm.types.llms.anthropic import (
     ANTHROPIC_BETA_HEADER_VALUES,
     ANTHROPIC_HOSTED_TOOLS,
+    ANTHROPIC_PROMPT_CACHING_SCOPE_BETA_HEADER,
 )
 from litellm.types.llms.anthropic_tool_search import get_tool_search_beta_header
 from litellm.types.llms.vertex_ai import VertexPartnerProvider
@@ -65,6 +66,10 @@ def validate_anthropic_messages_environment(
         if existing_beta:
             beta_values.update(b.strip() for b in existing_beta.split(","))
 
+        # Use the helper to remove unsupported beta headers
+        self.remove_unsupported_beta(headers)
+        beta_values.discard(ANTHROPIC_PROMPT_CACHING_SCOPE_BETA_HEADER)
+
         # Check for web search tool
         for tool in tools:
             if isinstance(tool, dict) and tool.get("type", "").startswith(ANTHROPIC_HOSTED_TOOLS.WEB_SEARCH.value):
@@ -123,3 +128,23 @@ def transform_anthropic_messages_request(
         )  # do not pass output_format in request body to vertex ai - vertex ai does not support output_format as yet
 
         return anthropic_messages_request
+
+    def remove_unsupported_beta(self, headers: dict) -> None:
+        """
+        Helper method to remove unsupported beta headers from the beta headers.
+        Modifies headers in place.
+        """
+        unsupported_beta_headers = [
+            ANTHROPIC_PROMPT_CACHING_SCOPE_BETA_HEADER
+        ]
+        existing_beta = headers.get("anthropic-beta")
+        if existing_beta:
+            filtered_beta = [
+                b.strip()
+                for b in existing_beta.split(",")
+                if b.strip() not in unsupported_beta_headers
+            ]
+            if filtered_beta:
+                headers["anthropic-beta"] = ",".join(filtered_beta)
+            elif "anthropic-beta" in headers:
+                del headers["anthropic-beta"]

litellm/types/llms/anthropic.py

@@ -647,4 +647,4 @@ class ANTHROPIC_BETA_HEADER_VALUES(str, Enum):
 ANTHROPIC_OAUTH_TOKEN_PREFIX = "sk-ant-oat"
 ANTHROPIC_OAUTH_BETA_HEADER = "oauth-2025-04-20"
 
-
+ANTHROPIC_PROMPT_CACHING_SCOPE_BETA_HEADER = "prompt-caching-scope-2026-01-05"

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

@@ -37,8 +37,9 @@ def test_get_supported_params_thinking():
 def test_vertex_ai_anthropic_web_search_header_in_completion():
     """Test that web search tool adds the required beta header for Vertex AI completion requests"""
     from unittest.mock import MagicMock, patch
+
     from litellm.llms.anthropic.common_utils import AnthropicModelInfo
-    
+
     # Create the config instance
     model_info = AnthropicModelInfo()
 
@@ -259,3 +260,33 @@ def test_vertex_ai_anthropic_other_models_still_use_tools():
     assert "tools" in result_params, "Claude 3 Sonnet should also use tool-based structured output"
     assert "tool_choice" in result_params, "Tool choice should be present"
     assert "json_mode" in result_params, "JSON mode should be enabled"
+
+def test_vertex_ai_partner_models_anthropic_remove_prompt_caching_scope_beta_header():
+    """
+    Test that remove_unsupported_beta correctly filters out prompt-caching-scope-2026-01-05 
+    from the anthropic-beta headers.
+    """
+    from litellm.llms.vertex_ai.vertex_ai_partner_models.anthropic.experimental_pass_through.transformation import (
+        VertexAIPartnerModelsAnthropicMessagesConfig,
+    )
+
+    # This beta header should be removed
+    PROMPT_CACHING_BETA_HEADER = "prompt-caching-scope-2026-01-05"
+    headers = {
+        "anthropic-beta": f"other-feature,{PROMPT_CACHING_BETA_HEADER},web-search-2025-03-05"
+    }
+
+    config = VertexAIPartnerModelsAnthropicMessagesConfig()
+    config.remove_unsupported_beta(headers)
+
+    beta_header = headers.get("anthropic-beta")
+    assert PROMPT_CACHING_BETA_HEADER not in (beta_header or ""), \
+        f"{PROMPT_CACHING_BETA_HEADER} should be filtered out"
+    assert "other-feature" in (beta_header or ""), \
+        "Other non-excluded beta headers should remain"
+    assert "web-search-2025-03-05" in (beta_header or ""), \
+        "Other non-excluded beta headers should remain"
+    # If prompt-caching was the only value, header should be removed completely
+    headers2 = {"anthropic-beta": PROMPT_CACHING_BETA_HEADER}
+    config.remove_unsupported_beta(headers2)
+    assert "anthropic-beta" not in headers2, "Header should be removed if no supported values remain"