AI
OpenAI releases GPT-Live-1 API with simultaneous listening and speaking
OpenAI has released GPT-Live-1 in its API, offering a voice model that listens and speaks simultaneously while delegating deeper reasoning and tool use to a separate backend model. Developers can customise conversational style, use telephony support and choose their reasoning infrastructure. The front-end voice layer costs $0.05 per minute. The architecture could simplify interruption handling in voice applications, although the reported performance improvements and customer results come from OpenAI’s announcement rather than independent testing.
Editorial summary · Publisher article
What matters
- Backend reasoning can use an OpenAI model or a third-party model.
- The API provides speech-recognition transcripts, response text, keyword biasing and explicit turn detection.
- OpenAI reports a 30-percentage-point improvement over GPT-Realtime-2.1 on Full Duplex Bench.
Context & caveats
- The quoted per-minute price covers the voice layer; it does not establish the total cost of a complete agent.
- Speak’s reported reduction in interruptions comes from early evaluations.
Why this made the edition
The announcement supplies substantive architecture, integration, pricing and availability details beyond launch promotion. Performance results remain company-reported.
Source material