fix(proxy): async_post_call_streaming_iterator_hook now properly iterates async generators

[rcII]

Summary

Fixes #9639

The async_post_call_streaming_iterator_hook function in litellm/proxy/utils.py was broken:

1. Was a sync function (def) not an async generator
2. Returned an AsyncGenerator without ever iterating it
3. Callback generators were chained but never consumed - meaning callbacks never actually processed any chunks

Changes

1. Changed def to async def making it an async generator
2. Added async for chunk in current_response: yield chunk to actually iterate through the callback chain
3. Added unit tests verifying the fix

The Bug (Before)

def async_post_call_streaming_iterator_hook(self, response, ...):  # NOT async!
    for callback in litellm.callbacks:
        response = _callback.async_post_call_streaming_iterator_hook(...)  # Assigns AsyncGenerator
    return response  # Returns UNCONSUMED generator - callbacks never execute!

The Fix (After)

async def async_post_call_streaming_iterator_hook(self, response, ...):  # Now async!
    current_response = response
    for callback in litellm.callbacks:
        current_response = _callback.async_post_call_streaming_iterator_hook(...)
    
    # Actually iterate and yield chunks
    async for chunk in current_response:
        yield chunk

Test Plan

• Added 4 unit tests in tests/test_litellm/proxy/hooks/test_async_post_call_streaming_iterator_hook.py
• Tests verify async generator behavior, callback chaining, empty callbacks, and error propagation
• All tests pass
• Black formatting applied
• Ruff linting passes

Related

• Issue: [Bug]: Cannot Modify Streaming Responses: async_post_call_streaming_iterator_hook Not Executed in Proxy #9639
• Prior fix attempt: kazimurtaza@a4a4ce9

[CLAassistant]

All committers have signed the CLA.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Preview	Comments	Updated (UTC)
litellm	Ready	Preview	Comment	Dec 7, 2025 5:59am