GPT‑Live‑1 hits the API at $0.05 per minute

GPT‑Live‑1 was first introduced in ChatGPT and is now being made available in the API, giving developers access to a voice model that listens and speaks at the same time. The release focuses on capabilities that let developers steer voice experiences around their users and workflows, such as interruption handling, reasoning and tool calling delegation, and tone, pace, and style control through the system prompt. It also improves long-session reliability and handles background noise and silence without interrupting the conversation.

The model simplifies voice-agent architecture by replacing traditional setups that stitch together speech-to-text, a reasoning model, and text-to-speech. Each handoff in those setups adds latency and creates opportunities to lose timing, context, or conversational rhythm. GPT‑Live‑1 handles listening and speaking in a single model, allowing it to respond to interruptions and acknowledgements as they happen while delegating deeper reasoning to the back end. Developers choose the models, tools, and agent harness behind the conversation, matching reasoning depth, speed, and cost to each task.

The outcome hinges on whether developers adopt the single-model approach for telephony and customer support, and whether custom voice access becomes broadly available. OpenAI says it will continue to expand voice options and language availability over the coming months, which may determine how quickly businesses can deploy voice agents that fit their product and sound natural to users.