OpenAI introduced GPT-Live to power ChatGPT Voice, with separate variants for different plans. The announcement emphasizes more natural interaction and describes limitations at launch, including differences in language fluency and the absence of video and screen sharing in the new mode.

Context

A conversational voice should be evaluated across an exchange, not just a prepared sentence. Does it handle pauses, corrections, and a change of topic? Product boundaries matter too: a newer voice model may improve speech while initially lacking a feature available in an older mode.

Sources & authors

  1. Introducing GPT-Live
    OpenAI · July 8, 2026
OpenAItools