OpenAI adds GPT-Live-1.1 and standby to its voice API

OpenAI added GPT-Live-1.1 to its Live API on November 19 and introduced a standby option for browser and app voice sessions that use WebRTC. The company’s Python library release added the model to its Live session choices the same day.
The model documentation says GPT-Live-1.1 can listen while it speaks and improves transcription and accent consistency across languages compared with GPT-Live-1. Developers can use the new model in an existing Live session configuration. The voice model can pass requests to a separate backend for reasoning or tool use.
Standby works with GPT-Live-1 as well as 1.1. It pauses inference during inactivity and resumes when someone speaks. OpenAI lists GPT-Live-1.1 at $0.05 per active minute and $0.02 per minute in standby, with separate charges for backend models and tools.
This is a developer API update, not the first appearance of the model: ChatGPT Voice began rolling it out in October. Standby gives developers a cheaper way to leave a session connected during a lull without paying the active rate.
Sources and citations
Revision history
Published .