OpenAI is updating ChatGPT's Voice Mode with new models, GPT-Live-1 and GPT-Live-1-mini, part of the GPT-Live family, built on a full-duplex architecture that allows them to speak and listen simultaneously. GPT-Live-1 replaces Advanced Voice Mode as the default voice option in ChatGPT, understanding long pauses and adding phrases like “mhmm,” “yeah,” or “got it” to show attention, moving away from the previous turn-based approach. According to OpenAI, GPT-Live-1 is the smartest voice model it has released to date; when a query requires research, reasoning, or agentic capabilities, it delegates the task to the frontier GPT-5.5 model while continuing to talk. Users can choose the level of smartness by tapping Settings, selecting the Intelligence button, and choosing “Instant for fast responses, or Medium and High when you want ChatGPT to spend more time thinking.” For topics like weather, stocks, and sports, rich visual cards may appear in the chat window. The GPT-Live models are currently rolling out to ChatGPT users worldwide on iOS, Android, and web. For Go, Plus, and Pro users, GPT-Live-1 will be the default voice model, while GPT‑Live‑1 mini will be the default for free users. Legacy models like Standard and Advanced Voice Mode remain accessible from the app.