OpenAI launches its next-generation speech model, GPT-Live.

OpenAI has launched a new voice interaction model, GPT-Live, which supports real-time simultaneous interpretation and full-duplex dialogue, allowing users to interrupt at any time. The model supports customizable inference strength and introduces a deep task delegation mechanism, enabling smooth voice interaction in the foreground while handling tasks such as online searches and complex inference in the background. Visual cards displaying weather, sports events, and other information can also be displayed in real-time during conversations.