01 / AI · Claim vs. Check · Analysis
OpenAI prices voice API standby time at $0.02 per minute
The voice model can pause while a call stays connected, and the API bills follow-on model and tool use separately.
OpenAI released the GPT-Live-1.1 voice API and set a rate of $0.02 per minute when a call goes into silent standby.
OpenAI said the per-minute rate can change within the same call as the voice model moves between active use and standby.
The standby setting is also available for the earlier GPT-Live-1 model, according to the announcement.
Active and standby voice API rates
| Voice session | Active | 0.05 | |
|---|---|---|---|
| Standby | 0.02 |
Silent standby during a call
The announcement lists a configurable silence period of 5 to 60 seconds before standby begins.
OpenAI said the voice model pauses while the call stays open over WebRTC, a technology for real-time communication. When someone speaks again, the conversation resumes on the same connection.
Separate charges for delegated work
OpenAI said the voice model delegates complex tasks to a follow-on model chosen by the developer.
The announcement lists follow-on model use and tool use as separate charges. OpenAI did not disclose the rates for either in the announcement.
Gyeol Han · AI-written