OpenAI launches GPT Live, a continuous voice interaction model
A full-duplex architecture lets the model listen and speak simultaneously and decide whether to interrupt, pause or hand off to GPT-5.5, replacing ChatGPT's turn-based voice mode.
- Models & capabilities
- Minor
OpenAI launched GPT Live, a new voice system for ChatGPT built on a full-duplex architecture that processes input and output at the same time, replacing the turn-based Advanced Voice Mode it succeeds. Where the previous system waited for one side of a conversation to finish before generating a response, GPT Live can listen and speak simultaneously, produce filler acknowledgements such as “mhmm” while a user is still talking, and decide — multiple times a second — whether to keep listening, speak, pause, interrupt or invoke a tool.
The system is built in two layers: a continuous-interaction layer that manages the moment-to-moment flow of conversation, and a delegation layer that hands more demanding requests to GPT-5.5 in the background so the voice exchange itself is not held up by slower reasoning. OpenAI rolled the models out globally on iOS, Android and the web the same day: GPT-Live-1 became the default for Go, Plus and Pro subscribers, with a smaller GPT-Live-1 mini serving Free-tier users.
The launch targeted ChatGPT’s existing voice and dictation user base, which OpenAI has said exceeds 150 million weekly users, and followed a broader pattern across the industry — including Google’s and Meta’s own real-time voice products — of treating latency and interruption-handling, rather than raw language quality, as the remaining obstacle to voice interfaces feeling conversational.