GPT-Live-1 reaches the API at $0.05 per minute for voice

01 What happened

On September 10, 2026, OpenAI made GPT-Live-1 available via API for full-duplex voice conversations that listen and speak simultaneously.

02 Key details

• The model manages interruptions and pauses natively, delegating deeper reasoning and tool usage to a developer-selected backend model.
• Developers can define tone, pace, and style using system instructions for voice agents, including automated phone workflows.
• The voice layer costs $0.05 per minute, while backend model and tool usage incur separate charges.
• OpenAI reports that early Speak evaluations demonstrated nearly 80% fewer interruptions during thinking pauses compared to previous turn-based systems.

03 Why it matters

This API allows developers to engineer responsive voice agents and phone workflows that maintain a natural conversational flow by handling interruptions and pauses in real-time.

04 Who it matters to

Software developers and voice system engineers.