usage_metadata in invoke: difference between Langchain and LangServe

[lorerave85]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

llm = ChatOpenAI()
resp = llm.invoke("what is ice cream?")
print(resp.usage_metadata)

> {'input_tokens': 11, 'output_tokens': 193, 'total_tokens': 204}

Description

I have a simple chain that returns me a lot of metadata using invoke, visible in the response.

response_metadata={'token_usage': {'completion_tokens': 193, 'prompt_tokens': 11, 'total_tokens': 204, 'completion_tokens_details': {'audio_tokens': None, 'reasoning_tokens': 0}, 'prompt_tokens_details': {'audio_tokens': None, 'cached_tokens': 0}}, 'model_name': 'gpt-4o-mini-2024-07-18', 'system_fingerprint': 'fp_482c22a7bc', 'finish_reason': 'stop', 'logprobs': None}, id='run-ac0e12e4-4a0f-403c-98f2-c7ff057924e8-0', usage_metadata={'input_tokens': 11, 'output_tokens': 193, 'total_tokens': 204}

if I try to run the same simple chain via langserve, the metadata reported is practically non-existent.

"metadata": {
"run_id": "f3529d30-1adc-4xxxxxttyy-yyyyy9b3d3e",
"feedback_tokens": []
}

is there a way to bring back some metadata also in the langserve response?

i don't want to reimplement a custom callback, i see that on langsmith all this metadata arrives.

Thanks
Lorenzo

System Info

langchain==0.3.3
langchain-chroma==0.1.4
langchain-cli==0.0.21
langchain-community==0.3.2
langchain-core==0.3.10
langchain-experimental==0.0.49
langchain-openai==0.2.0
langchain-text-splitters==0.3.0