OpenAI has announced the availability of GPT-Live-1 in its API, a model designed to power more natural voice experiences for developers building conversational applications.
According to the company, GPT-Live-1 supports full-duplex voice conversations, meaning interactions can flow in both directions rather than following a rigid turn-taking pattern. OpenAI says the model also offers stronger instruction following, the ability to use custom voices, and support for telephony.
The release targets developers who want to embed real-time, voice-driven interactions into their products through the API. No pricing, availability dates, or benchmark figures were included in the announcement.