OpenAI launches GPT-Live-1: one model for listening and speaking with tunable tone

OpenAIDevs · x · 2026-09-11

OpenAI's developer account detailed GPT-Live-1, a voice model that handles listening and speaking in a single model, cutting model handoffs for faster conversation. Developers can control tone, pacing, expressiveness, response length and language via instructions; the model mirrors the speaker's tone and emotion. Users can keep talking while a backend model handles reasoning and tool calls.

Related event: OpenAI launches full-duplex speech model GPT-Live-1 on API(26 posts)→

Original post →

More from Models

Models channel →