OpenAI launches GPT-Live for more natural voice conversations

Business Insider reports that OpenAI introduced GPT-Live during a July 8 livestream. The new voice model supports full-duplex conversation, meaning ChatGPT can listen while speaking and handle interruptions more naturally. OpenAI also demonstrated live translation and multitask planning, positioning GPT-Live as a step toward less rigid audio interfaces.

Key details: GPT-Live supports full-duplex voice interaction, OpenAI demonstrated live translation and fluid multi-step conversations, The model is part of a broader push toward continuous AI interfaces.

Why it matters: Voice is becoming a frontier-model interface again, with labs competing on interruption handling, translation, and continuous conversation instead of only text benchmarks.