OpenAI Rebuilds ChatGPT Voice With GPT-Live to Talk and Listen at Once
OpenAI's GPT-Live gets a full engineering teardown: WARP protocol, stateful inference, and async delegation make voice feel truly live
- GPT-Live goes full-duplex: OpenAI's new voice system listens and speaks simultaneously, removing the turn detector that caused unnatural interruptions.
- Two-model architecture: A lightweight voice model handles conversation flow while GPT-5.5 handles search and reasoning asynchronously in the background.
- WARP protocol: OpenAI built and open-sourced a new WebRTC extension that cuts session startup from 6 network round trips to 1, now submitted to the IETF.
- Stateful inference with seamless handoffs: Context compaction and model instance transitions happen without pausing the audio stream.
- Availability: GPT-Live-1 mini is free; full GPT-Live-1 requires Go ($8/mo) or above. API access is not yet available.
- No video or screen sharing at launch: Legacy Advanced Voice Mode remains available for those features while GPT-Live catches up.
OpenAI just published a detailed engineering post on how they rebuilt ChatGPT Voice from the ground up to power GPT-Live, their third-generation voice system. The headline feature is full-duplex audio -- the model can listen and speak at the same time -- but the deeper story is a complete rearchitecture of every layer between the user's microphone and the model, built over six months to make that feel natural at 150 million weekly voice users.
Why the old approach hit a wall
Every previous ChatGPT voice system was fundamentally turn-based. Advanced Voice Mode processed and generated audio within a single model, reducing latency and making conversations smoother -- but it still operated through discrete turns. The model had to wait for the user to stop speaking before responding, resulting in rigid back-and-forth. The system relied on a small "turn detector" model to decide when the user had finished talking. Because turn detection was based on silence, even a brief pause or background noise could be mistaken for the end of a turn -- causing the model to interrupt at unnatural times.
This marks the third generation of ChatGPT's voice technology in two years, with OpenAI aiming to create a smoother user experience that feels more natural than simply querying a search engine.
The two architectural bets
GPT-Live is built on two structural changes that work together:
- Full-duplex continuous inference: Instead of processing a sequence of separate messages, GPT-Live continuously processes input while generating output. The model can therefore make interaction decisions many times per second: whether to speak, continue listening, pause, interrupt, or invoke a tool.
- Asynchronous delegation: GPT-Live introduces a structural change that decouples the voice interaction layer from the reasoning layer. When a user asks a straightforward question, GPT-Live handles it directly. But when the query demands web search, deeper reasoning, or more complex agentic work, GPT-Live delegates the task to a frontier model running in the background -- at launch, GPT-5.5 -- and continues talking with the user while the computation happens asynchronously.
The practical result: that continuous-processing design enables live translation and lets the model acknowledge a speaker mid-sentence with cues like "mhmm," rather than waiting for a full turn to end.