OpenAI Releases GPT-Live-1.1 in API, Cuts Idle Voice Session Costs by 60%
OpenAI announced on 19 November 2026 that its GPT-Live-1.1 voice model is available in the API from 20 November, alongside a Standby mode for idle sessions. The model is the same one powering voice in ChatGPT Plus and Pro, and it runs as a full-duplex voice API that handles interruptions. Standby mode pauses inference after a configurable 5 to 60 second silence while keeping the WebRTC call open, reducing the idle rate from $0.05 to $0.02 per minute. Standby works with both GPT-Live-1 and GPT-Live-1.1, and billing is per second, though backend model and tool usage are charged separately. Developers should note that GPT-Live runs on a different endpoint than the Realtime API, so migrating from Realtime is not a simple model swap.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in