diff --git a/litellm/llms/anthropic/chat/transformation.py b/litellm/llms/anthropic/chat/transformation.py
index 86378b97d2e..7c48474a663 100644
--- a/litellm/llms/anthropic/chat/transformation.py
+++ b/litellm/llms/anthropic/chat/transformation.py
@@ -200,6 +200,71 @@ class AnthropicConfig(AnthropicModelInfo, BaseConfig):
 
         return params
 
+    @staticmethod
+    def filter_anthropic_output_schema(schema: Dict[str, Any]) -> Dict[str, Any]:
+        """
+        Filter out unsupported fields from JSON schema for Anthropic's output_format API.
+        
+        Anthropic's output_format doesn't support certain JSON schema properties:
+        - maxItems: Not supported for array types
+        - minItems: Not supported for array types
+        
+        This function recursively removes these unsupported fields while preserving
+        all other valid schema properties.
+        
+        Args:
+            schema: The JSON schema dictionary to filter
+            
+        Returns:
+            A new dictionary with unsupported fields removed
+            
+        Related issue: https://github.com/BerriAI/litellm/issues/19444
+        """
+        if not isinstance(schema, dict):
+            return schema
+        
+        # Fields that Anthropic doesn't support in output_format
+        unsupported_fields = {"maxItems", "minItems"}
+        
+        result = {}
+        for key, value in schema.items():
+            # Skip unsupported fields
+            if key in unsupported_fields:
+                continue
+            
+            # Recursively filter nested structures
+            if key == "properties" and isinstance(value, dict):
+                result[key] = {
+                    k: AnthropicConfig.filter_anthropic_output_schema(v)
+                    for k, v in value.items()
+                }
+            elif key == "items" and isinstance(value, dict):
+                result[key] = AnthropicConfig.filter_anthropic_output_schema(value)
+            elif key == "$defs" and isinstance(value, dict):
+                result[key] = {
+                    k: AnthropicConfig.filter_anthropic_output_schema(v)
+                    for k, v in value.items()
+                }
+            elif key == "anyOf" and isinstance(value, list):
+                result[key] = [
+                    AnthropicConfig.filter_anthropic_output_schema(item)
+                    for item in value
+                ]
+            elif key == "allOf" and isinstance(value, list):
+                result[key] = [
+                    AnthropicConfig.filter_anthropic_output_schema(item)
+                    for item in value
+                ]
+            elif key == "oneOf" and isinstance(value, list):
+                result[key] = [
+                    AnthropicConfig.filter_anthropic_output_schema(item)
+                    for item in value
+                ]
+            else:
+                result[key] = value
+        
+        return result
+
     def get_json_schema_from_pydantic_object(
         self, response_format: Union[Any, Dict, None]
     ) -> Optional[dict]:
@@ -636,9 +701,13 @@ class AnthropicConfig(AnthropicModelInfo, BaseConfig):
         )
         if json_schema is None:
             return None
+        
+        # Filter out unsupported fields for Anthropic's output_format API
+        filtered_schema = self.filter_anthropic_output_schema(json_schema)
+        
         return AnthropicOutputSchema(
             type="json_schema",
-            schema=json_schema,
+            schema=filtered_schema,
         )
 
     def map_response_format_to_anthropic_tool(
diff --git a/tests/test_litellm/llms/anthropic/test_anthropic_structured_output.py b/tests/test_litellm/llms/anthropic/test_anthropic_structured_output.py
new file mode 100644
index 00000000000..7f6b68881d5
--- /dev/null
+++ b/tests/test_litellm/llms/anthropic/test_anthropic_structured_output.py
@@ -0,0 +1,169 @@
+"""
+Test Anthropic structured output with Pydantic models.
+
+This test file verifies that Pydantic models with various constraints
+are properly converted to Anthropic-compatible JSON schemas.
+"""
+
+import pytest
+from pydantic import BaseModel, Field
+from typing import List
+
+
+class TestAnthropicStructuredOutput:
+    """Test Anthropic structured output schema transformations."""
+
+    def test_max_length_on_list_field_filtered(self):
+        """
+        Test that max_length on List fields is filtered out for Anthropic models.
+        
+        Anthropic doesn't support 'maxItems' property for array types in their
+        output_format.schema, so we need to filter it out.
+        
+        Related issue: https://github.com/BerriAI/litellm/issues/19444
+        """
+        from litellm.llms.anthropic.chat.transformation import AnthropicConfig
+        
+        # Define a Pydantic model with max_length on a List field
+        class ResponseModel(BaseModel):
+            items: List[str] = Field(max_length=5, description="List of items")
+            name: str = Field(description="Name field")
+        
+        config = AnthropicConfig()
+        
+        # Get the JSON schema from the Pydantic model
+        json_schema = config.get_json_schema_from_pydantic_object(ResponseModel)
+        
+        # Extract the actual schema
+        schema = json_schema["json_schema"]["schema"]
+        
+        # Verify that maxItems is present in the raw schema (from Pydantic)
+        assert "maxItems" in schema["properties"]["items"]
+        
+        # Now apply the Anthropic output format transformation
+        response_format = {
+            "type": "json_schema",
+            "json_schema": json_schema["json_schema"]
+        }
+        
+        output_format = config.map_response_format_to_anthropic_output_format(
+            response_format
+        )
+        
+        # Verify that maxItems is filtered out for Anthropic
+        assert output_format is not None
+        assert "schema" in output_format
+        transformed_schema = output_format["schema"]
+        
+        # maxItems should be removed from the items property
+        assert "maxItems" not in transformed_schema["properties"]["items"]
+        
+        # But other properties should remain
+        assert "type" in transformed_schema["properties"]["items"]
+        assert transformed_schema["properties"]["items"]["type"] == "array"
+        assert "description" in transformed_schema["properties"]["items"]
+
+    def test_min_length_on_list_field_filtered(self):
+        """
+        Test that min_length on List fields is filtered out for Anthropic models.
+        
+        Anthropic likely doesn't support 'minItems' either.
+        """
+        from litellm.llms.anthropic.chat.transformation import AnthropicConfig
+        
+        class ResponseModel(BaseModel):
+            items: List[str] = Field(min_length=2, description="List of items")
+        
+        config = AnthropicConfig()
+        json_schema = config.get_json_schema_from_pydantic_object(ResponseModel)
+        
+        response_format = {
+            "type": "json_schema",
+            "json_schema": json_schema["json_schema"]
+        }
+        
+        output_format = config.map_response_format_to_anthropic_output_format(
+            response_format
+        )
+        
+        assert output_format is not None
+        transformed_schema = output_format["schema"]
+        
+        # minItems should be removed
+        assert "minItems" not in transformed_schema["properties"]["items"]
+
+    def test_nested_array_constraints_filtered(self):
+        """
+        Test that array constraints are filtered at all nesting levels.
+        """
+        from litellm.llms.anthropic.chat.transformation import AnthropicConfig
+        
+        class NestedItem(BaseModel):
+            tags: List[str] = Field(max_length=3)
+        
+        class ResponseModel(BaseModel):
+            items: List[NestedItem] = Field(max_length=5)
+        
+        config = AnthropicConfig()
+        json_schema = config.get_json_schema_from_pydantic_object(ResponseModel)
+        
+        response_format = {
+            "type": "json_schema",
+            "json_schema": json_schema["json_schema"]
+        }
+        
+        output_format = config.map_response_format_to_anthropic_output_format(
+            response_format
+        )
+        
+        assert output_format is not None
+        transformed_schema = output_format["schema"]
+        
+        # Top-level maxItems should be removed
+        assert "maxItems" not in transformed_schema["properties"]["items"]
+        
+        # Nested maxItems should also be removed
+        if "$defs" in transformed_schema:
+            nested_item_schema = transformed_schema["$defs"].get("NestedItem", {})
+            if "properties" in nested_item_schema and "tags" in nested_item_schema["properties"]:
+                assert "maxItems" not in nested_item_schema["properties"]["tags"]
+
+    def test_other_constraints_preserved(self):
+        """
+        Test that other valid constraints are preserved.
+        """
+        from litellm.llms.anthropic.chat.transformation import AnthropicConfig
+        
+        class ResponseModel(BaseModel):
+            name: str = Field(max_length=100, min_length=1, description="Name")
+            age: int = Field(ge=0, le=150, description="Age")
+            items: List[str] = Field(description="Items list")
+        
+        config = AnthropicConfig()
+        json_schema = config.get_json_schema_from_pydantic_object(ResponseModel)
+        
+        response_format = {
+            "type": "json_schema",
+            "json_schema": json_schema["json_schema"]
+        }
+        
+        output_format = config.map_response_format_to_anthropic_output_format(
+            response_format
+        )
+        
+        assert output_format is not None
+        transformed_schema = output_format["schema"]
+        
+        # String constraints should be preserved
+        name_schema = transformed_schema["properties"]["name"]
+        assert "maxLength" in name_schema
+        assert "minLength" in name_schema
+        assert name_schema["maxLength"] == 100
+        assert name_schema["minLength"] == 1
+        
+        # Number constraints should be preserved
+        age_schema = transformed_schema["properties"]["age"]
+        assert "minimum" in age_schema
+        assert "maximum" in age_schema
+        assert age_schema["minimum"] == 0
+        assert age_schema["maximum"] == 150