ChatGPT Live could make talking to AI feel straight out of the movies

ChatGPT Live aims to make voice conversations with AI feel more natural, closer to sci‑fi movies, by updating how ChatGPT Voice talks and listens. OpenAI is rolling out GPT‑Live, built on a full‑duplex architecture that lets the model process user speech and generate its own replies at the same time, choosing when to speak, pause, or keep listening. OpenAI says it handles common conversational interruptions—like asking a follow‑up, slowing down, or staying quiet—more smoothly, including brief acknowledgments such as “mhmm” or “got it.” The system also is designed to focus better on a user’s voice in noisy settings like traffic. GPT‑Live can continue the spoken exchange while delegating tougher tasks to GPT‑5.5. It launches globally on iOS, Android, and the web, with GPT‑Live‑1 for Plus/Pro/Go users and GPT‑Live‑1 mini for free. Limitations remain around video and screen sharing, which OpenAI says will arrive later.