OpenAI Releases New Voice Models for More Natural Live Conversations
TechCrunch, Wednesday, July 8th, 2026
OpenAI launches GPT-Live-1 voice models with full-duplex speech for more natural live conversations.
OpenAI introduced GPT-Live-1 and GPT-Live-1 mini, conversational models with full-duplex technology that lets them speak and listen simultaneously for natural interruptions.
They replace ChatGPT's Advanced Voice Mode, with the mini as default for all users and the larger model for paid subscribers, and can access GPT-5.5 for complex queries and live translation.
The technology supports extended 30-40 minute sessions. OpenAI frames voice as a future primary computing interface for long-running tasks but denies positioning it as an AI companion and adds age-appropriate safeguards.
Limitations remain, including unnatural accents in non-English languages seen in a Hindi translation demo.