OpenAI is retiring ChatGPT's old voice modes and moving every user to GPT-Live
Anyone still on the Advanced or Standard voices will be switched automatically to GPT-Live, the model launched in July that listens and talks at the same time during a conversation.
OpenAI is pulling the Advanced and Standard voice modes from ChatGPT starting December 11 this year. Anyone still using either option will be moved automatically to Live mode, which runs on GPT-Live, and the two older voices will no longer be available to pick from the voice menu in settings.
The company announced the change on Tuesday, according to The Tribune, which ran an ANI wire report on November 11. GPT-Live was introduced in July as OpenAI's newest voice model, built to make talking with ChatGPT feel more natural.
According to the company, voice usage has nearly doubled since the launch, and virtually everyone who talks to ChatGPT by voice has already moved to GPT-Live. "This change will allow us to simplify the voice experience and focus our efforts on improving GPT-Live," OpenAI said.
How GPT-Live works
The new model listens and speaks at the same time. That makes the back and forth faster, because users don't have to wait for the system to finish listening before they get an answer. During a conversation, it can react with short expressions, reply right away or stay quiet when the user needs a moment to think.
For more complex questions and tasks, GPT-Live connects in the background to OpenAI's advanced text models. The spoken conversation keeps going while the system works on the heavier request.
The old modes were built differently. Advanced used a model that processed and generated audio, but it waited for the user to stop talking before responding. Standard relied on three separate models: one turned speech into text, another generated the answer and a third converted that answer back into audio.
What stays available
Conversations held in Advanced and Standard modes remain in ChatGPT's history and can be picked up again by text or voice after the shutdown. Live video and screen sharing will keep running on Advanced until GPT-Live gets those features, and voice conversations with custom GPTs use Advanced until December 11.
OpenAI says it will use feedback from people who relied on the old voices to improve the new mode, especially the depth and detail of its answers, continuity across conversations and personalization.
For brand teams, what matters is the size of the channel: voice usage in ChatGPT has nearly doubled since July, according to OpenAI, and every spoken conversation will now run through a single mode that taps the company's text models when a question gets more complex.
Key questions
When do the Advanced and Standard voice modes stop working in ChatGPT?
Starting December 11 this year. After that date, users are moved automatically to Live mode, which runs on GPT-Live.
Will past voice conversations be lost?
No. They stay in ChatGPT's history and can be resumed by text or voice.
What's different about GPT-Live compared with the old voices?
It listens and speaks at the same time, without waiting for the user to finish before answering, and it connects in the background to OpenAI's advanced text models for complex requests.
Source: The Tribune.