OpenAI has announced a new advanced voice ai model, with reports varying on the specific name and version, including gpt-realtime, gpt-live-1, and gpt-realtime-2.1. The model is described as having full-duplex capabilities, allowing it to listen and speak simultaneously, and is intended for use in low-latency voice agents. The new model is available in OpenAI's API, with updates aimed at advancing voice intelligence and enabling more natural ai voice conversations.