[Bug]: `server_tool_use` missing in usage when using `stream=True`

[KeremTurgutlu]

What happened?

Usage field server_tool_use is missing when using streaming.

For example when running the following code:

from litellm import completion 

o = completion(
  model="claude-sonnet-4-5-20250929",
  messages=[
    {"role": "user", "content": "What is the weather today in NYC?"},
  ],
  web_search_options = {"search_context_size": "medium"}
)
print(o)

usage metadata includes web_search_requets.

Usage(completion_tokens=326, prompt_tokens=9457, total_tokens=9783, completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=0, text_tokens=None, image_tokens=None), server_tool_use=ServerToolUse(web_search_requests=1), cache_creation_input_tokens=0, cache_read_input_tokens=0)

But this is not the case when streaming is enabled, it only includes WebSearchResultBlock:

o = completion(
  model="claude-sonnet-4-5-20250929",
  messages=[
    {"role": "user", "content": "What is the weather today in NYC?"},
  ],
  web_search_options = {"search_context_size": "medium"},
  stream=True,
  stream_options={"include_usage": True}
)

for chunk in o: print(chunk)

('usage', Usage(completion_tokens=287, prompt_tokens=9463, total_tokens=9750, completion_tokens_details=CompletionTokensDetailsWrapper(accepted_prediction_tokens=None, audio_tokens=None, reasoning_tokens=0, rejected_prediction_tokens=None, text_tokens=None), prompt_tokens_details=None)

Relevant log output

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

1.76.0

Twitter / LinkedIn details

No response