OpenAI GPT-Live: ChatGPT's New Full-Duplex Voice AI

OpenAI has completely rebuilt ChatGPT's voice mode. The new system, called GPT-Live, allows the AI to listen and talk at the same time. This is a big change from how voice AI used to work. Previously, your speech was transcribed, processed, and then converted back into speech sequentially. Now, GPT-Live processes incoming audio while simultaneously generating its own speech. This means the AI can make decisions multiple times a second, like when to keep talking or when to listen. It can even offer small acknowledgments, similar to human conversation. For complex tasks, GPT-Live delegates to OpenAI's Frontier text model, GPT 5.5, to maintain conversation flow. This new design aims to make interacting with AI feel more like talking to another person.