pipecat/changelog/4217.changed.md

• When multiple function calls are returned in a single LLM response, by default (when group_parallel_tools=True) the LLM is now triggered exactly once after the last call in the batch completes, rather than waiting for all function calls.