OpenAI is making GPT-Live-1 available to developers through an API, bringing full-duplex voice conversations to apps that can listen and speak at the same time. The model costs $0.05 per minute, billed by the second, while backend model and tool use are charged separately.
GPT-Live-1 API is built for interruptions
GPT-Live-1 continuously processes incoming audio while generating a response. It can decide whether to keep listening, pause, interrupt, speak or call a tool several times per second, rather than waiting for a neat turn to end.
The model can also hand deeper reasoning and tool work to a separate backend agent. OpenAI says GPT-Live-1 can keep the conversation moving while another model handles search or a more complex task. The API documentation lists gpt-live-1 as a full-duplex model with audio and text input and output, plus function calling and streaming.
The official model documentation lists a Live Sessions endpoint at v1/live/sessions. GPT-Live-1 is not supported through the standard Chat Completions, Responses or Realtime endpoints.
GPT-Live-1 API pricing and performance
The API price is $0.05 per minute, with sessions billed per second rather than rounded up to a full minute. Backend Responses calls and tool usage use the normal pricing for the configured model and tools.
According to The Decoder, OpenAI’s benchmarks put GPT-Live-1 at 80.1 percent in full-duplex interactivity, compared with 45.4 percent for GPT-Realtime-2.1. Turn-taking latency falls to 0.8 seconds from 1.4 seconds, while tool-calling accuracy rises to 87 percent from 60 percent.
The model includes 12 voices covering different accents, dialects and languages. It also returns automatic speech-recognition transcripts and response text, giving developers a text record alongside the audio conversation.
Where developers could use GPT-Live-1
Yelp is using GPT-Live-1 for phone-based reservations and service-business communications. OpenAI says developers can pair the voice layer with a backend model and tools chosen for the job, so the same conversation system can sit in front of different business logic.
That could include reservation lines, appointment booking and customer support for UAE businesses. OpenAI has not announced UAE-specific pricing, hosting or a local rollout for GPT-Live-1, so any deployment here would depend on the developer’s account, data requirements and service design.
GPT-Live-1 is a more capable voice interface, but the per-minute charge is only the first line on the bill. Backend reasoning, tool calls and the engineering needed to keep a phone conversation useful will decide whether it works as a business product.
Internal coverage: OpenAI’s UAE inference residency is live and Meta AI’s desktop app adds voice features.
What is GPT-Live-1?
GPT-Live-1 is OpenAI’s full-duplex voice model. It can listen and speak at the same time and hand deeper reasoning or tool use to a backend agent.
How much does the GPT-Live-1 API cost?
Voice sessions cost $0.05 per minute, billed by the second. Backend model and tool usage are charged separately.
Which endpoint does GPT-Live-1 use?
OpenAI’s documentation lists the Live Sessions endpoint as v1/live/sessions.
