fix(responses): streaming with tool_choice allowed_tools fails

[Chesars]

Title

fix: streaming with tool_choice allowed_tools fails with pickle error

Relevant issues

Fixes #17192

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard
  requirement - see details
• I have added a screenshot of my new test passing locally
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem

Type

🐛 Bug Fix

Changes

Summary

When using the Responses API with stream=True and tool_choice of type allowed_tools, the streaming fails with:
TypeError: cannot pickle 'pydantic_core._pydantic_core.ValidatorIterator' object. In _handle_logging_completed_response() when calling copy.deepcopy() on the completed response. The tool_choice field contains a Pydantic ValidatorIterator that cannot be serialized with pickle.

Solution

Replace copy.deepcopy() with Pydantic's model_dump() + model_validate() pattern in both async and sync _handle_logging_completed_response() methods.

Files Changed

• litellm/responses/streaming_iterator.py
• tests/llm_responses_api_testing/test_base_responses_api_streaming_iterator.py

Test passing

======================== 6 passed, 9 warnings in 0.09s =========================

[vercel]

@Chesars is attempting to deploy a commit to the CLERKIEAI Team on Vercel.

A member of the Team first needs to authorize it.