Thanks for reaching out! When you're working with OpenAI's realtime model, the

llm_node

doesn't get triggered. Instead, the interaction with the LLM happens directly through the Realtime API connection and its event stream, which means it skips the usual STT-LLM-TTS pipeline flow. Now, if you're using a non-realtime LLM with a pipeline agent, the

llm_node

does get triggered once per user turn—right after the speech-to-text processing finishes and just before the text-to-speech synthesis kicks in. Sources: Pipeline nodes and hooks | LiveKit Docs | Pipeline nodes and hooks | LiveKit Docs | Docs::Agents | livekit-plugins/livekit-plugins-openai/livekit/plugins/openai/realtime/real... | agent.py

When you're working with OpenAI's realtime model, the

llm_node

doesn't get triggered. Instead, the interaction with the LLM happens smoothly through the Realtime API connection and its event stream, so it skips the usual STT-LLM-TTS pipeline flow. If you're using a non-realtime LLM with a pipeline agent, then the

llm_node

will be triggered once per user turn—right after the speech-to-text processing finishes and just before the text-to-speech synthesis starts.