Fix Anthropic streaming web search usage

Co-authored-by: ishaan-berri <ishaan-berri@users.noreply.github.com>

litellm/types/utils.py

@@ -1553,7 +1553,7 @@ class Usage(SafeAttributeModel, CompletionUsage):
         completion_tokens_details: Optional[
             Union[CompletionTokensDetailsWrapper, dict]
         ] = None,
-        server_tool_use: Optional[ServerToolUse] = None,
+        server_tool_use: Optional[Union[ServerToolUse, Dict[str, Any]]] = None,
         cost: Optional[float] = None,
         **params,
     ):
@@ -1655,6 +1655,8 @@ class Usage(SafeAttributeModel, CompletionUsage):
         )
 
         if server_tool_use is not None:
+            if isinstance(server_tool_use, dict):
+                server_tool_use = ServerToolUse(**server_tool_use)
             self.server_tool_use = server_tool_use
         else:  # maintain openai compatibility in usage object if possible
             del self.server_tool_use

tests/test_litellm/litellm_core_utils/test_streaming_chunk_builder_utils.py

@@ -8,7 +8,7 @@ sys.path.insert(
     0, os.path.abspath("../../..")
 )  # Adds the parent directory to the system path
 
-from litellm import stream_chunk_builder
+from litellm import completion_cost, stream_chunk_builder
 from litellm.litellm_core_utils.streaming_chunk_builder_utils import ChunkProcessor
 from litellm.types.utils import (
     ChatCompletionDeltaToolCall,
@@ -520,7 +520,66 @@ def test_stream_chunk_builder_anthropic_web_search():
     assert usage.prompt_tokens == 50
     assert usage.completion_tokens == 27
     assert usage.total_tokens == 77
-    assert usage.server_tool_use["web_search_requests"] == 2
+    assert usage.server_tool_use.web_search_requests == 2
+
+
+def test_stream_chunk_builder_anthropic_web_search_cost_calculation():
+    chunk1 = ModelResponseStream(
+        id="chatcmpl-mocked-usage-1",
+        created=1745513206,
+        model="claude-sonnet-4-5-20250929",
+        object="chat.completion.chunk",
+        system_fingerprint=None,
+        choices=[
+            StreamingChoices(
+                finish_reason=None,
+                index=0,
+                delta=Delta(content="", role="assistant"),
+                logprobs=None,
+            )
+        ],
+        provider_specific_fields=None,
+        stream_options={"include_usage": True},
+        usage=Usage(
+            completion_tokens=0,
+            prompt_tokens=50,
+            total_tokens=50,
+            completion_tokens_details=None,
+            server_tool_use=ServerToolUse(web_search_requests=2),
+            prompt_tokens_details=None,
+        ),
+    )
+
+    chunk2 = ModelResponseStream(
+        id="chatcmpl-mocked-usage-1",
+        created=1745513207,
+        model="claude-sonnet-4-5-20250929",
+        object="chat.completion.chunk",
+        system_fingerprint=None,
+        choices=[
+            StreamingChoices(
+                finish_reason="stop",
+                index=0,
+                delta=Delta(content="done", role=None),
+                logprobs=None,
+            )
+        ],
+        provider_specific_fields=None,
+        stream_options={"include_usage": True},
+        usage=Usage(
+            completion_tokens=27,
+            prompt_tokens=0,
+            total_tokens=27,
+            completion_tokens_details=None,
+            prompt_tokens_details=None,
+        ),
+    )
+
+    response = stream_chunk_builder(chunks=[chunk1, chunk2])
+
+    assert response is not None
+    assert isinstance(response.usage.server_tool_use, ServerToolUse)
+    assert completion_cost(completion_response=response) > 0
 
 
 def test_sort_chunks_handles_dict_hidden_params_created_at():

tests/test_litellm/types/test_types_utils.py

@@ -45,6 +45,7 @@ def test_usage_dump():
     from litellm.types.utils import (
         CompletionTokensDetailsWrapper,
         PromptTokensDetailsWrapper,
+        ServerToolUse,
         Usage,
     )
 
@@ -74,6 +75,16 @@ def test_usage_dump():
     new_usage = Usage(**current_usage.model_dump())
     assert new_usage.prompt_tokens_details.web_search_requests == 1
 
+    server_tool_usage = Usage(
+        completion_tokens=10,
+        prompt_tokens=5,
+        total_tokens=15,
+        server_tool_use=ServerToolUse(web_search_requests=2),
+    )
+    new_server_tool_usage = Usage(**server_tool_usage.model_dump())
+    assert isinstance(new_server_tool_usage.server_tool_use, ServerToolUse)
+    assert new_server_tool_usage.server_tool_use.web_search_requests == 2
+
 
 def test_usage_completion_tokens_details_text_tokens():
     from litellm.types.utils import Usage