🇺🇸 English 🇰🇷 한국어 🇯🇵 日本語
Home › Developer Tools
Developer Tools

GPT-Live-1.1 Hits API With 2-Cent Standby

OpenAI released GPT-Live-1.1 in its API on Nov. 19 and added a standby mode that bills idle WebRTC voice sessions at $0.02 a minute instead of $0.05, for both GPT-Live-1 and 1.1.

OpenAI has released GPT-Live-1.1 in its API and added a standby mode that bills idle voice sessions at 2 cents a minute instead of 5. The company listed both changes in its API changelog on Thursday, Nov. 19, giving developers a newer full-duplex voice model and a cheaper way to keep a voice line open while nobody is speaking.

Two women on a London street, each looking at or talking on a mobile phone
Voice agents often sit on open lines with long silences, which standby is meant to price lower. File photo. 사진: garryknight / Flickr(CC BY 2.0)

Key takeaways

  • GPT-Live-1.1 is now available through the Live API as gpt-live-1.1, with better pronunciation, accent consistency and transcription in multilingual conversations, according to OpenAI.
  • A new standby mode pauses inference on WebRTC sessions after 5 to 60 seconds without speech and bills that time at $0.02 a minute, 60% below the $0.05 active rate.
  • Both prices apply to GPT-Live-1 and GPT-Live-1.1; backend model and tool usage are still billed separately.

What GPT-Live-1.1 changes for developers

GPT-Live is OpenAI's full-duplex voice family, meaning it listens while it speaks instead of waiting for a turn. GPT-Live-1 became generally available in the API on Sept. 10 at 5 cents a minute. ChatGPT's own voice mode moved to version 1.1 on Oct. 15, but OpenAI had not said then when developers would get it.

On its model page, OpenAI says GPT-Live-1.1 "transcribes speech more accurately and maintains a more consistent accent and voice across languages" than GPT-Live-1. Switching requires only changing the model name in an existing Live session configuration. The model accepts and returns audio and text, supports function calling and does not support structured outputs. OpenAI did not publish benchmark scores for the new version.

The two models carry separate concurrent-session limits, so moving traffic to 1.1 does not eat into a team's GPT-Live-1 quota.

How GPT-Live standby pricing works

Standby is configured when a WebRTC session is created. Setting session.wake_policy.type to "speech" tells the session to wait for a voice before starting inference. A second setting, session.idle_timeout_s, takes a whole number from 5 to 60 and returns the session to standby after that many seconds without an input or output transcript.

The connection and session ID stay the same, and the session restores its conversation history when someone speaks again. A voice activity threshold from 0 to 1, defaulting to 0.5, controls how much evidence of speech is needed to wake it. OpenAI warns that the first reply after standby "may take longer."

GPT-Live session pricing (OpenAI pricing page, Nov. 21, 2026)
ModelActiveStandby
gpt-live-1.1$0.05 / min$0.02 / min
gpt-live-1$0.05 / min$0.02 / min

Sessions are billed per second. As an illustration, a 10-minute session that sits idle for six of those minutes would cost 32 cents in voice time with standby, against 50 cents without it.

Context: voice agents and idle time

Idle time is a real cost for always-on assistants, in-car voice and support lines where callers pause to look something up. OpenAI's guide still recommends closing a session entirely during long gaps and starting a new one later. Rivals are pushing on price too, including Google's new voice models that talk while they think.

GPT-Live has expanded steadily this fall. OpenAI added nine natural voices on Oct. 16, including Korean and Japanese options, and opened outbound SIP phone calling to all organizations on Nov. 2.

What's next

The changelog describes standby only for WebRTC sessions; OpenAI has not said whether WebSocket or SIP connections will get it. Independent tests of version 1.1's accent and transcription claims have yet to appear.

FAQ

How much does GPT-Live-1.1 cost? $0.05 a minute while active and $0.02 a minute in standby, billed per second, plus charges for any backend model or tools.

Do I need to rewrite my GPT-Live-1 app? No. OpenAI says developers can use gpt-live-1.1 in an existing Live session configuration.

Does standby work on GPT-Live-1 too? Yes. OpenAI says standby is available for both GPT-Live-1 and GPT-Live-1.1 WebRTC sessions.

References

#OpenAI #GPT-Live #GPT-Live-1.1 #Voice AI #AI Agents #Live API #WebRTC #API Pricing
S
Seung Jung
AI Newsway · Developer

Editor responsible for AI Newsway. Writes about AI, developer tools, cloud and productivity technology.

More by this reporter ›
Copyright © AI Newsway All rights reserved. Unauthorized reproduction, redistribution, and AI training prohibited.
Share
f X K

Comments 0

Comments are currently disabled.

Be the first to comment.