OpenAI opens GPT-Live-1.1 to API with $0.02/min standby
TL;DR
- OpenAI has made GPT-Live-1.1, its full-duplex voice model, available through the API under the identifier gpt-live-1.1.
- A new standby configuration bills idle time at $0.02 per minute versus $0.05 per minute for active inference, with configurable 5 to 60 second timeouts.
- The same model has powered ChatGPT Plus and Pro voice since October 15, with highlighted upgrades to cross-language accents and letter and number recognition.
OpenAI's GPT-Live-1.1 voice model is now available through the API, with a new standby configuration that bills idle time at $0.02 per minute compared with $0.05 per minute for active inference, the company wrote on its developer community forum.
The standby option, available for both GPT-Live-1 and GPT-Live-1.1, lets developers set idle timeouts between 5 and 60 seconds. When silence exceeds the threshold, the model pauses but the WebRTC call stays open, and conversation automatically resumes when speech returns. Backend model and tool usage is billed separately on top.
The same model has powered ChatGPT Plus and Pro voice since October 15. OpenAI's post credits it with "more natural and consistent accents across languages" and precise understanding of letters and numbers, alongside simultaneous listening and speaking and the ability to delegate workflows to backend models without interrupting the call. The API drop ships 38 voice IDs, with seven new entries including calandria, hornero and luciernaga that lean Spanish-language. API access uses the model identifier gpt-live-1.1. It lands a day after OpenAI's ChatGPT Live Dictation launch, the second voice-side ship we've tracked from the company in two days.
Originally reported by openai.com
Read the original article →Original headline: OpenAI Puts GPT-Live-1.1 in API With 60% Cheaper Standby Mode at $0.02 per Minute