**[OpenAI Launches GPT-Live-1 API, Supporting Full-Duplex Voice Conversations]** OpenAI has announced the release of the GPT-Live-1 full-duplex voice model in its API, capable of simultaneously listening and generating speech. This improves handling of interruptions, pauses, background noise, and extended conversations, and supports telephone scenarios. Developers can control tone, speech rate, and conversational style through system prompts, and delegate complex reasoning and tool usage to backend text models like GPT-6 Astra or third-party models. OpenAI stated that GPT-Live-1 outperforms GPT-Realtime-2.1 by 30 percentage points on the Full Duplex Bench. Early testing by the language learning platform Speak shows that the model reduces misinterpretation of learners' thoughtful pauses by nearly 80% compared to traditional turn-based systems.