GPT-Live 1
OpenAI · Audio model
GPT-Live 1 is OpenAI's full-duplex voice model, launched on 8 July 2026 to replace Advanced Voice Mode in ChatGPT. It can keep listening while it speaks and delegates complex requests to a stronger model in the background. Developers can use…
Current
ReleasedJul 8, 2026
Context window—
Input → outputaudio, text → audio, text
API price$0.05 per minute (billed per second; backend Responses calls billed at normal model rates)
Consumer accessPowers ChatGPT Voice: GPT-Live-1 is the default on paid ChatGPT plans, GPT-Live-1 mini on the free tier (iOS, Android, web)
Best for
Natural, full-duplex voice conversations where the assistant can listen and talk at the same time.
Overview
GPT-Live 1 is OpenAI's full-duplex voice model, launched on 8 July 2026 to replace Advanced Voice Mode in ChatGPT. It can keep listening while it speaks and delegates complex requests to a stronger model in the background. Developers can use it through a separate Live API endpoint at $0.05 per minute.
Strengths and limits
- Full-duplex: listens and speaks simultaneously, handles interruptions and pauses
- Hands harder reasoning and tool use off to a backend model while continuing to talk
- Simple per-minute API pricing via a dedicated Live endpoint (v1/live/sessions)
- Only the Live endpoint is supported (not Realtime, Chat Completions or Responses)
- No image or video input; voice with video/screen share was not supported in ChatGPT at launch