OpenAI introduces gpt-live-1 for enhanced voice interactions

OpenAI has launched GPT-Live-1 via its API, offering developers a more natural and responsive voice model for building voice-enabled applications and business workflows. The model, first introduced in ChatGPT, enables full-duplex conversations by allowing simultaneous listening and speaking, with the ability to delegate complex reasoning to paired models and tools. Early evaluations showed a 80% reduction in interruptions compared to traditional turn-based systems. GPT-Live-1 improves Full Duplex Bench performance by 30 percentage points over GPT-Realtime-2.1 and ranks #1 on Tau3 when paired with GPT-6 Astra. Developers can customize voice experiences with support for multiple accents, dialects, and languages. The API is available at $0.05 per minute for the front-end voice layer.