OpenAI introduces three new real-time voice models, including a GPT-5-level reasoning engine, to make voice the primary AI interface.
