OpenAI is giving ChatGPT Voice mode a major upgrade this week, alongside the expected release of GPT-5.6. The new experience runs on GPT-Live, a fresh voice model built to make conversations with ChatGPT feel more natural, faster, and easier to interrupt when needed.
OpenAI announced the new voice mode during a livestream featuring Kundan Kumar, Yuchen Zhang, Ehsan Asdar, and Rithesh Kumar. The company also shared more details about GPT-Live, which powers the updated voice experience across ChatGPT.
ChatGPT Voice Gets a More Natural Upgrade
The biggest change comes from GPT-Live’s full-duplex design, which lets ChatGPT listen and speak at the same time. This means users can talk over ChatGPT while it is still responding, and the assistant can adjust without forcing the conversation to stop and restart.
The new voice mode also understands pauses better, so it can stay quiet when users need more time to think. It can also respond with small listening cues such as “mhmm” or “yeah,” which should make longer conversations feel less stiff.
OpenAI says GPT-Live can also use web search, giving voice mode better access to current information when users need fresh answers.
Two GPT-Live Models Are Rolling Out
OpenAI is launching GPT-Live with two models: GPT-Live-1 and GPT-Live-1-mini. Free users will get GPT-Live-1-mini by default, while paid users across all tiers will get GPT-Live-1.
The rollout covers ChatGPT on iOS, Android, and the web. However, GPT-Live does not support voice with video or screen sharing at launch, so users still need legacy ChatGPT Voice mode for those features.
OpenAI says it has optimized GPT-Live for popular ChatGPT languages, though some languages may still have accent or fluency gaps for now.
