GPT-Live-1.1 pricing: how active and standby minutes are billed

OpenAI bills GPT-Live-1.1 voice sessions by time rather than tokens, charging $0.05 per active minute and $0.02 per standby minute, calculated per second without rounding up. Backend model and tool calls are billed separately at their usual rates. Active time covers speaking, silence before the idle timeout expires, and waiting for backend processing, while standby applies only to WebRTC sessions when inference is paused. Each WebRTC session also incurs a 15-second setup charge that is refunded once the session runs. Developers can control costs by setting a speech wake policy and an idle timeout between 5 and 60 seconds, since omitting the timeout keeps inference active and prevents standby pricing from applying.
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