This commit fixes issue #4056 where token usage always returns 0 when using
async streaming crew kickoff.
Root cause: The streaming completion methods (_handle_streaming_completion and
_ahandle_streaming_completion) in OpenAICompletion never called
_track_token_usage_internal(), unlike the non-streaming methods.
Changes:
- Add stream_options={'include_usage': True} to streaming params so OpenAI API
returns usage information in the final chunk
- Extract and track token usage from the final chunk in sync streaming
- Extract and track token usage from the final chunk in async streaming
- Extract and track token usage from final_completion in response_model paths
- Add _extract_chunk_token_usage method for ChatCompletionChunk objects
- Add tests to verify streaming token usage tracking works correctly
Co-Authored-By: João <joao@crewai.com>
558fc6eda4