client.runs.stream hangs or loses real-time updates when using url=None (In-process Transport) in v0.6

[kissycn]

Checked other resources

• This is a bug, not a usage question. For questions, please use the LangChain Forum (https://forum.langchain.com/).
• I added a clear and detailed title that summarizes the issue.
• I read what a minimal reproducible example is (https://stackoverflow.com/help/minimal-reproducible-example).
• I included a self-contained, minimal example that demonstrates the issue INCLUDING all the relevant imports. The code run AS IS to reproduce the issue.

Example Code

# This setup causes the issue since in v0.6
client = get_client(url=None)

# This works fine and provides real-time updates
client = get_client(url="http://localhost:2024")

async for chunk in client.runs.stream(
    thread_id=thread_id,
    assistant_id="my_agent",
    input=input_data,
    stream_mode="messages" # or "messages-tuple"
):
    print(chunk) # No output until the run finishes

Error Message and Stack Trace (if applicable)

The stream either:
Hangs indefinitely.
Buffers all output and returns everything at once only after the run is completed.

Description

After upgrading from LangGraph API 0.5 to 0.6 , the streaming functionality (client.runs.stream) no longer provides real-time updates when using the in-process connection (get_client(url=None)).

I am running the client and the LangGraph server within the same FastAPI process.
I suspect the issue might be related to how the new ASGITransport handles streaming or the default msgpack serialization in v0.6.
In v0.5, this exact setup (url=None) worked without any buffering issues.
I've tried passing headers={"Accept": "application/json"} but the issue persists with url=None.

System Info

LangGraph API Version: 0.6.18
LangGraph SDK Version: 0.4.11
Python Version: 3.12
Operating System: MacOS

[hinthornw]

This is only in dev? Or also with up/ in prod?