OpenAI launches GPT-Live voice model
GPT-Live-1 and GPT-Live-1 mini can listen and speak at once, backchanneling like a person, while handing hard queries to GPT-5.5 in the background.
OpenAI shipped GPT-Live on July 8, replacing ChatGPT’s existing Advanced and Standard Voice Modes with a new full-duplex model family. Full-duplex means the model can listen and speak simultaneously, so it can backchannel with a quiet “mhmm” while you’re still talking, or get interrupted mid-sentence without losing the thread, behavior closer to a real conversation than the turn-based voice modes that came before it.
Two versions are rolling out today: GPT-Live-1, the default for Go, Plus, and Pro subscribers, and a smaller GPT-Live-1 mini, the default for Free users. When a query needs real reasoning or a web search, GPT-Live quietly hands it off to GPT-5.5 in the background and keeps the conversation flowing while it waits on the answer, rather than going silent.
The rollout covers ChatGPT on web, iOS, and Android globally, starting today. API access for developers who want to build GPT-Live into their own products is “coming soon,” per OpenAI, with no date given yet. Business, Enterprise, and Education plans aren’t included in this rollout; video and screen-share still route through the older Advanced Voice Mode for now.
For teams building voice interfaces on top of OpenAI’s API, this is a preview of where the underlying model is headed rather than something to build against today, worth watching the changelog for the API release before planning around it.